libata-eh.c 112 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0-or-later
2 3 4 5 6 7
/*
 *  libata-eh.c - libata error handling
 *
 *  Copyright 2006 Tejun Heo <htejun@gmail.com>
 *
 *  libata documentation is available via 'make {ps|pdf}docs',
8
 *  as Documentation/driver-api/libata.rst
9 10 11 12 13 14
 *
 *  Hardware documentation available from http://www.t13.org/ and
 *  http://www.sata-io.org/
 */

#include <linux/kernel.h>
15
#include <linux/blkdev.h>
16
#include <linux/export.h>
17
#include <linux/pci.h>
18 19 20 21 22
#include <scsi/scsi.h>
#include <scsi/scsi_host.h>
#include <scsi/scsi_eh.h>
#include <scsi/scsi_device.h>
#include <scsi/scsi_cmnd.h>
23
#include <scsi/scsi_dbg.h>
Jeff Garzik's avatar
Jeff Garzik committed
24
#include "../scsi/scsi_transport_api.h"
25 26 27

#include <linux/libata.h>

Hannes Reinecke's avatar
Hannes Reinecke committed
28
#include <trace/events/libata.h>
29 30
#include "libata.h"

31
enum {
32
	/* speed down verdicts */
33 34 35
	ATA_EH_SPDN_NCQ_OFF		= (1 << 0),
	ATA_EH_SPDN_SPEED_DOWN		= (1 << 1),
	ATA_EH_SPDN_FALLBACK_TO_PIO	= (1 << 2),
36
	ATA_EH_SPDN_KEEP_ERRORS		= (1 << 3),
37 38 39

	/* error flags */
	ATA_EFLAG_IS_IO			= (1 << 0),
40
	ATA_EFLAG_DUBIOUS_XFER		= (1 << 1),
41
	ATA_EFLAG_OLD_ER                = (1 << 31),
42 43 44 45 46 47

	/* error categories */
	ATA_ECAT_NONE			= 0,
	ATA_ECAT_ATA_BUS		= 1,
	ATA_ECAT_TOUT_HSM		= 2,
	ATA_ECAT_UNK_DEV		= 3,
48 49 50 51 52
	ATA_ECAT_DUBIOUS_NONE		= 4,
	ATA_ECAT_DUBIOUS_ATA_BUS	= 5,
	ATA_ECAT_DUBIOUS_TOUT_HSM	= 6,
	ATA_ECAT_DUBIOUS_UNK_DEV	= 7,
	ATA_ECAT_NR			= 8,
53

54 55
	ATA_EH_CMD_DFL_TIMEOUT		=  5000,

56 57 58
	/* always put at least this amount of time between resets */
	ATA_EH_RESET_COOL_DOWN		=  5000,

59 60 61 62 63 64 65
	/* Waiting in ->prereset can never be reliable.  It's
	 * sometimes nice to wait there but it can't be depended upon;
	 * otherwise, we wouldn't be resetting.  Just give it enough
	 * time for most drives to spin up.
	 */
	ATA_EH_PRERESET_TIMEOUT		= 10000,
	ATA_EH_FASTDRAIN_INTERVAL	=  3000,
66 67

	ATA_EH_UA_TRIES			= 5,
68 69 70 71

	/* probe speed down parameters, see ata_eh_schedule_probe() */
	ATA_EH_PROBE_TRIAL_INTERVAL	= 60000,	/* 1 min */
	ATA_EH_PROBE_TRIALS		= 2,
72 73 74 75 76 77
};

/* The following table determines how we sequence resets.  Each entry
 * represents timeout for that try.  The first try can be soft or
 * hardreset.  All others are hardreset if available.  In most cases
 * the first reset w/ 10sec timeout should succeed.  Following entries
Dan Williams's avatar
Dan Williams committed
78 79
 * are mostly for error handling, hotplug and those outlier devices that
 * take an exceptionally long time to recover from reset.
80
 */
81
static const unsigned int ata_eh_reset_timeouts[] = {
82 83
	10000,	/* most drives spin up by 10sec */
	10000,	/* > 99% working drives spin up before 20sec */
Dan Williams's avatar
Dan Williams committed
84
	35000,	/* give > 30 secs of idleness for outlier devices */
85
	 5000,	/* and sweet one last chance */
86
	UINT_MAX, /* > 1 min has elapsed, give up */
87 88
};

89
static const unsigned int ata_eh_identify_timeouts[] = {
90 91 92
	 5000,	/* covers > 99% of successes and not too boring on failures */
	10000,  /* combined time till here is enough even for media access */
	30000,	/* for true idiots */
93
	UINT_MAX,
94 95
};

96
static const unsigned int ata_eh_revalidate_timeouts[] = {
97 98
	15000,	/* Some drives are slow to read log pages when waking-up */
	15000,  /* combined time till here is enough even for media access */
99
	UINT_MAX,
100 101
};

102
static const unsigned int ata_eh_flush_timeouts[] = {
103 104 105
	15000,	/* be generous with flush */
	15000,  /* ditto */
	30000,	/* and even more generous */
106
	UINT_MAX,
107 108
};

109
static const unsigned int ata_eh_other_timeouts[] = {
110 111 112
	 5000,	/* same rationale as identify timeout */
	10000,	/* ditto */
	/* but no merciful 30sec for other commands, it just isn't worth it */
113
	UINT_MAX,
114 115 116 117
};

struct ata_eh_cmd_timeout_ent {
	const u8		*commands;
118
	const unsigned int	*timeouts;
119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137
};

/* The following table determines timeouts to use for EH internal
 * commands.  Each table entry is a command class and matches the
 * commands the entry applies to and the timeout table to use.
 *
 * On the retry after a command timed out, the next timeout value from
 * the table is used.  If the table doesn't contain further entries,
 * the last value is used.
 *
 * ehc->cmd_timeout_idx keeps track of which timeout to use per
 * command class, so if SET_FEATURES times out on the first try, the
 * next try will use the second timeout value only for that class.
 */
#define CMDS(cmds...)	(const u8 []){ cmds, 0 }
static const struct ata_eh_cmd_timeout_ent
ata_eh_cmd_timeout_table[ATA_EH_CMD_TIMEOUT_TABLE_SIZE] = {
	{ .commands = CMDS(ATA_CMD_ID_ATA, ATA_CMD_ID_ATAPI),
	  .timeouts = ata_eh_identify_timeouts, },
138 139
	{ .commands = CMDS(ATA_CMD_READ_LOG_EXT, ATA_CMD_READ_LOG_DMA_EXT),
	  .timeouts = ata_eh_revalidate_timeouts, },
140 141 142 143 144 145 146 147
	{ .commands = CMDS(ATA_CMD_READ_NATIVE_MAX, ATA_CMD_READ_NATIVE_MAX_EXT),
	  .timeouts = ata_eh_other_timeouts, },
	{ .commands = CMDS(ATA_CMD_SET_MAX, ATA_CMD_SET_MAX_EXT),
	  .timeouts = ata_eh_other_timeouts, },
	{ .commands = CMDS(ATA_CMD_SET_FEATURES),
	  .timeouts = ata_eh_other_timeouts, },
	{ .commands = CMDS(ATA_CMD_INIT_DEV_PARAMS),
	  .timeouts = ata_eh_other_timeouts, },
148 149
	{ .commands = CMDS(ATA_CMD_FLUSH, ATA_CMD_FLUSH_EXT),
	  .timeouts = ata_eh_flush_timeouts },
150 151
	{ .commands = CMDS(ATA_CMD_VERIFY),
	  .timeouts = ata_eh_reset_timeouts },
152 153 154
};
#undef CMDS

155
static void __ata_port_freeze(struct ata_port *ap);
156 157
static int ata_eh_set_lpm(struct ata_link *link, enum ata_lpm_policy policy,
			  struct ata_device **r_failed_dev);
158
#ifdef CONFIG_PM
159 160
static void ata_eh_handle_port_suspend(struct ata_port *ap);
static void ata_eh_handle_port_resume(struct ata_port *ap);
161 162 163 164 165 166 167
#else /* CONFIG_PM */
static void ata_eh_handle_port_suspend(struct ata_port *ap)
{ }

static void ata_eh_handle_port_resume(struct ata_port *ap)
{ }
#endif /* CONFIG_PM */
168

169 170
static __printf(2, 0) void __ata_ehi_pushv_desc(struct ata_eh_info *ehi,
				 const char *fmt, va_list args)
171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194
{
	ehi->desc_len += vscnprintf(ehi->desc + ehi->desc_len,
				     ATA_EH_DESC_LEN - ehi->desc_len,
				     fmt, args);
}

/**
 *	__ata_ehi_push_desc - push error description without adding separator
 *	@ehi: target EHI
 *	@fmt: printf format string
 *
 *	Format string according to @fmt and append it to @ehi->desc.
 *
 *	LOCKING:
 *	spin_lock_irqsave(host lock)
 */
void __ata_ehi_push_desc(struct ata_eh_info *ehi, const char *fmt, ...)
{
	va_list args;

	va_start(args, fmt);
	__ata_ehi_pushv_desc(ehi, fmt, args);
	va_end(args);
}
195
EXPORT_SYMBOL_GPL(__ata_ehi_push_desc);
196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218

/**
 *	ata_ehi_push_desc - push error description with separator
 *	@ehi: target EHI
 *	@fmt: printf format string
 *
 *	Format string according to @fmt and append it to @ehi->desc.
 *	If @ehi->desc is not empty, ", " is added in-between.
 *
 *	LOCKING:
 *	spin_lock_irqsave(host lock)
 */
void ata_ehi_push_desc(struct ata_eh_info *ehi, const char *fmt, ...)
{
	va_list args;

	if (ehi->desc_len)
		__ata_ehi_push_desc(ehi, ", ");

	va_start(args, fmt);
	__ata_ehi_pushv_desc(ehi, fmt, args);
	va_end(args);
}
219
EXPORT_SYMBOL_GPL(ata_ehi_push_desc);
220 221 222 223 224 225 226 227 228 229 230 231 232 233 234

/**
 *	ata_ehi_clear_desc - clean error description
 *	@ehi: target EHI
 *
 *	Clear @ehi->desc.
 *
 *	LOCKING:
 *	spin_lock_irqsave(host lock)
 */
void ata_ehi_clear_desc(struct ata_eh_info *ehi)
{
	ehi->desc[0] = '\0';
	ehi->desc_len = 0;
}
235
EXPORT_SYMBOL_GPL(ata_ehi_clear_desc);
236

237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262
/**
 *	ata_port_desc - append port description
 *	@ap: target ATA port
 *	@fmt: printf format string
 *
 *	Format string according to @fmt and append it to port
 *	description.  If port description is not empty, " " is added
 *	in-between.  This function is to be used while initializing
 *	ata_host.  The description is printed on host registration.
 *
 *	LOCKING:
 *	None.
 */
void ata_port_desc(struct ata_port *ap, const char *fmt, ...)
{
	va_list args;

	WARN_ON(!(ap->pflags & ATA_PFLAG_INITIALIZING));

	if (ap->link.eh_info.desc_len)
		__ata_ehi_push_desc(&ap->link.eh_info, " ");

	va_start(args, fmt);
	__ata_ehi_pushv_desc(&ap->link.eh_info, fmt, args);
	va_end(args);
}
263
EXPORT_SYMBOL_GPL(ata_port_desc);
264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298

#ifdef CONFIG_PCI
/**
 *	ata_port_pbar_desc - append PCI BAR description
 *	@ap: target ATA port
 *	@bar: target PCI BAR
 *	@offset: offset into PCI BAR
 *	@name: name of the area
 *
 *	If @offset is negative, this function formats a string which
 *	contains the name, address, size and type of the BAR and
 *	appends it to the port description.  If @offset is zero or
 *	positive, only name and offsetted address is appended.
 *
 *	LOCKING:
 *	None.
 */
void ata_port_pbar_desc(struct ata_port *ap, int bar, ssize_t offset,
			const char *name)
{
	struct pci_dev *pdev = to_pci_dev(ap->host->dev);
	char *type = "";
	unsigned long long start, len;

	if (pci_resource_flags(pdev, bar) & IORESOURCE_MEM)
		type = "m";
	else if (pci_resource_flags(pdev, bar) & IORESOURCE_IO)
		type = "i";

	start = (unsigned long long)pci_resource_start(pdev, bar);
	len = (unsigned long long)pci_resource_len(pdev, bar);

	if (offset < 0)
		ata_port_desc(ap, "%s %s%llu@0x%llx", name, type, len, start);
	else
299 300
		ata_port_desc(ap, "%s 0x%llx", name,
				start + (unsigned long long)offset);
301
}
302
EXPORT_SYMBOL_GPL(ata_port_pbar_desc);
303 304
#endif /* CONFIG_PCI */

305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332
static int ata_lookup_timeout_table(u8 cmd)
{
	int i;

	for (i = 0; i < ATA_EH_CMD_TIMEOUT_TABLE_SIZE; i++) {
		const u8 *cur;

		for (cur = ata_eh_cmd_timeout_table[i].commands; *cur; cur++)
			if (*cur == cmd)
				return i;
	}

	return -1;
}

/**
 *	ata_internal_cmd_timeout - determine timeout for an internal command
 *	@dev: target device
 *	@cmd: internal command to be issued
 *
 *	Determine timeout for internal command @cmd for @dev.
 *
 *	LOCKING:
 *	EH context.
 *
 *	RETURNS:
 *	Determined timeout.
 */
333
unsigned int ata_internal_cmd_timeout(struct ata_device *dev, u8 cmd)
334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367
{
	struct ata_eh_context *ehc = &dev->link->eh_context;
	int ent = ata_lookup_timeout_table(cmd);
	int idx;

	if (ent < 0)
		return ATA_EH_CMD_DFL_TIMEOUT;

	idx = ehc->cmd_timeout_idx[dev->devno][ent];
	return ata_eh_cmd_timeout_table[ent].timeouts[idx];
}

/**
 *	ata_internal_cmd_timed_out - notification for internal command timeout
 *	@dev: target device
 *	@cmd: internal command which timed out
 *
 *	Notify EH that internal command @cmd for @dev timed out.  This
 *	function should be called only for commands whose timeouts are
 *	determined using ata_internal_cmd_timeout().
 *
 *	LOCKING:
 *	EH context.
 */
void ata_internal_cmd_timed_out(struct ata_device *dev, u8 cmd)
{
	struct ata_eh_context *ehc = &dev->link->eh_context;
	int ent = ata_lookup_timeout_table(cmd);
	int idx;

	if (ent < 0)
		return;

	idx = ehc->cmd_timeout_idx[dev->devno][ent];
368
	if (ata_eh_cmd_timeout_table[ent].timeouts[idx + 1] != UINT_MAX)
369 370 371
		ehc->cmd_timeout_idx[dev->devno][ent]++;
}

372
static void ata_ering_record(struct ata_ering *ering, unsigned int eflags,
373 374 375 376 377 378 379 380 381 382
			     unsigned int err_mask)
{
	struct ata_ering_entry *ent;

	WARN_ON(!err_mask);

	ering->cursor++;
	ering->cursor %= ATA_ERING_SIZE;

	ent = &ering->ring[ering->cursor];
383
	ent->eflags = eflags;
384 385 386 387
	ent->err_mask = err_mask;
	ent->timestamp = get_jiffies_64();
}

388 389 390 391 392 393 394 395 396
static struct ata_ering_entry *ata_ering_top(struct ata_ering *ering)
{
	struct ata_ering_entry *ent = &ering->ring[ering->cursor];

	if (ent->err_mask)
		return ent;
	return NULL;
}

397 398 399
int ata_ering_map(struct ata_ering *ering,
		  int (*map_fn)(struct ata_ering_entry *, void *),
		  void *arg)
400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417
{
	int idx, rc = 0;
	struct ata_ering_entry *ent;

	idx = ering->cursor;
	do {
		ent = &ering->ring[idx];
		if (!ent->err_mask)
			break;
		rc = map_fn(ent, arg);
		if (rc)
			break;
		idx = (idx - 1 + ATA_ERING_SIZE) % ATA_ERING_SIZE;
	} while (idx != ering->cursor);

	return rc;
}

418
static int ata_ering_clear_cb(struct ata_ering_entry *ent, void *void_arg)
419 420 421 422 423 424 425 426 427 428
{
	ent->eflags |= ATA_EFLAG_OLD_ER;
	return 0;
}

static void ata_ering_clear(struct ata_ering *ering)
{
	ata_ering_map(ering, ata_ering_clear_cb, NULL);
}

429 430
static unsigned int ata_eh_dev_action(struct ata_device *dev)
{
Tejun Heo's avatar
Tejun Heo committed
431
	struct ata_eh_context *ehc = &dev->link->eh_context;
432 433 434 435

	return ehc->i.action | ehc->i.dev_action[dev->devno];
}

436
static void ata_eh_clear_action(struct ata_link *link, struct ata_device *dev,
437 438
				struct ata_eh_info *ehi, unsigned int action)
{
439
	struct ata_device *tdev;
440 441 442

	if (!dev) {
		ehi->action &= ~action;
Tejun Heo's avatar
Tejun Heo committed
443
		ata_for_each_dev(tdev, link, ALL)
444
			ehi->dev_action[tdev->devno] &= ~action;
445 446 447 448 449 450
	} else {
		/* doesn't make sense for port-wide EH actions */
		WARN_ON(!(action & ATA_EH_PERDEV_MASK));

		/* break ehi->action into ehi->dev_action */
		if (ehi->action & action) {
Tejun Heo's avatar
Tejun Heo committed
451
			ata_for_each_dev(tdev, link, ALL)
452 453
				ehi->dev_action[tdev->devno] |=
					ehi->action & action;
454 455 456 457 458 459 460 461
			ehi->action &= ~action;
		}

		/* turn off the specified per-dev action */
		ehi->dev_action[dev->devno] &= ~action;
	}
}

462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496
/**
 *	ata_eh_acquire - acquire EH ownership
 *	@ap: ATA port to acquire EH ownership for
 *
 *	Acquire EH ownership for @ap.  This is the basic exclusion
 *	mechanism for ports sharing a host.  Only one port hanging off
 *	the same host can claim the ownership of EH.
 *
 *	LOCKING:
 *	EH context.
 */
void ata_eh_acquire(struct ata_port *ap)
{
	mutex_lock(&ap->host->eh_mutex);
	WARN_ON_ONCE(ap->host->eh_owner);
	ap->host->eh_owner = current;
}

/**
 *	ata_eh_release - release EH ownership
 *	@ap: ATA port to release EH ownership for
 *
 *	Release EH ownership for @ap if the caller.  The caller must
 *	have acquired EH ownership using ata_eh_acquire() previously.
 *
 *	LOCKING:
 *	EH context.
 */
void ata_eh_release(struct ata_port *ap)
{
	WARN_ON_ONCE(ap->host->eh_owner != current);
	ap->host->eh_owner = NULL;
	mutex_unlock(&ap->host->eh_mutex);
}

497 498 499 500 501 502
static void ata_eh_dev_disable(struct ata_device *dev)
{
	ata_acpi_on_disable(dev);
	ata_down_xfermask_limit(dev, ATA_DNXFER_FORCE_PIO0 | ATA_DNXFER_QUIET);
	dev->class++;

503 504
	/*
	 * From now till the next successful probe, ering is used to
505 506 507
	 * track probe failures.  Clear accumulated device error info.
	 */
	ata_ering_clear(&dev->ering);
508 509

	ata_dev_free_resources(dev);
510 511
}

512 513 514 515 516 517
static void ata_eh_unload(struct ata_port *ap)
{
	struct ata_link *link;
	struct ata_device *dev;
	unsigned long flags;

518 519 520 521 522 523 524 525 526 527 528 529 530
	/*
	 * Unless we are restarting, transition all enabled devices to
	 * standby power mode.
	 */
	if (system_state != SYSTEM_RESTART) {
		ata_for_each_link(link, ap, PMP_FIRST) {
			ata_for_each_dev(dev, link, ENABLED)
				ata_dev_power_set_standby(dev);
		}
	}

	/*
	 * Restore SControl IPM and SPD for the next driver and
531 532 533 534
	 * disable attached devices.
	 */
	ata_for_each_link(link, ap, PMP_FIRST) {
		sata_scr_write(link, SCR_CONTROL, link->saved_scontrol & 0xff0);
535 536
		ata_for_each_dev(dev, link, ENABLED)
			ata_eh_dev_disable(dev);
537 538 539 540 541 542 543 544 545 546 547 548
	}

	/* freeze and set UNLOADED */
	spin_lock_irqsave(ap->lock, flags);

	ata_port_freeze(ap);			/* won't be thawed */
	ap->pflags &= ~ATA_PFLAG_EH_PENDING;	/* clear pending from freeze */
	ap->pflags |= ATA_PFLAG_UNLOADED;

	spin_unlock_irqrestore(ap->lock, flags);
}

549 550 551 552 553 554 555 556 557 558 559 560
/**
 *	ata_scsi_error - SCSI layer error handler callback
 *	@host: SCSI host on which error occurred
 *
 *	Handles SCSI-layer-thrown error events.
 *
 *	LOCKING:
 *	Inherited from SCSI layer (none, can sleep)
 *
 *	RETURNS:
 *	Zero.
 */
561
void ata_scsi_error(struct Scsi_Host *host)
562
{
563
	struct ata_port *ap = ata_shost_to_port(host);
564
	unsigned long flags;
James Bottomley's avatar
James Bottomley committed
565
	LIST_HEAD(eh_work_q);
566

James Bottomley's avatar
James Bottomley committed
567 568 569 570
	spin_lock_irqsave(host->host_lock, flags);
	list_splice_init(&host->eh_cmd_q, &eh_work_q);
	spin_unlock_irqrestore(host->host_lock, flags);

571 572 573 574 575 576 577
	ata_scsi_cmd_error_handler(host, ap, &eh_work_q);

	/* If we timed raced normal completion and there is nothing to
	   recover nr_timedout == 0 why exactly are we doing error recovery ? */
	ata_scsi_port_error_handler(host, ap);

	/* finish or retry handled scmd's and clean up */
578
	WARN_ON(!list_empty(&eh_work_q));
579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596

}

/**
 * ata_scsi_cmd_error_handler - error callback for a list of commands
 * @host:	scsi host containing the port
 * @ap:		ATA port within the host
 * @eh_work_q:	list of commands to process
 *
 * process the given list of commands and return those finished to the
 * ap->eh_done_q.  This function is the first part of the libata error
 * handler which processes a given list of failed commands.
 */
void ata_scsi_cmd_error_handler(struct Scsi_Host *host, struct ata_port *ap,
				struct list_head *eh_work_q)
{
	int i;
	unsigned long flags;
597 598
	struct scsi_cmnd *scmd, *tmp;
	int nr_timedout = 0;
599

600 601
	/* make sure sff pio task is not running */
	ata_sff_flush_pio_task(ap);
602

Jeff Garzik's avatar
Jeff Garzik committed
603
	/* synchronize with host lock and sort out timeouts */
604

605
	/*
606
	 * For EH, all qcs are finished in one of three ways -
607
	 * normal completion, error completion, and SCSI timeout.
Alan Cox's avatar
Alan Cox committed
608
	 * Both completions can race against SCSI timeout.  When normal
609
	 * completion wins, the qc never reaches EH.  When error
610
	 * completion wins, the qc has ATA_QCFLAG_EH set.
611 612 613 614 615 616 617
	 *
	 * When SCSI timeout wins, things are a bit more complex.
	 * Normal or error completion can occur after the timeout but
	 * before this point.  In such cases, both types of
	 * completions are honored.  A scmd is determined to have
	 * timed out iff its associated qc is active and not failed.
	 */
618
	spin_lock_irqsave(ap->lock, flags);
619

620 621 622 623 624 625 626 627 628 629 630 631
	/*
	 * This must occur under the ap->lock as we don't want
	 * a polled recovery to race the real interrupt handler
	 *
	 * The lost_interrupt handler checks for any completed but
	 * non-notified command and completes much like an IRQ handler.
	 *
	 * We then fall into the error recovery code which will treat
	 * this as if normal completion won the race
	 */
	if (ap->ops->lost_interrupt)
		ap->ops->lost_interrupt(ap);
632

633 634
	list_for_each_entry_safe(scmd, tmp, eh_work_q, eh_entry) {
		struct ata_queued_cmd *qc;
635

636 637 638 639 640 641 642 643
		/*
		 * If the scmd was added to EH, via ata_qc_schedule_eh() ->
		 * scsi_timeout() -> scsi_eh_scmd_add(), scsi_timeout() will
		 * have set DID_TIME_OUT (since libata does not have an abort
		 * handler). Thus, to clear DID_TIME_OUT, clear the host byte.
		 */
		set_host_byte(scmd, DID_OK);

644 645 646 647 648
		ata_qc_for_each_raw(ap, qc, i) {
			if (qc->flags & ATA_QCFLAG_ACTIVE &&
			    qc->scsicmd == scmd)
				break;
		}
649

650 651
		if (i < ATA_MAX_QUEUE) {
			/* the scmd has an associated qc */
652
			if (!(qc->flags & ATA_QCFLAG_EH)) {
653 654
				/* which hasn't failed yet, timeout */
				qc->err_mask |= AC_ERR_TIMEOUT;
655
				qc->flags |= ATA_QCFLAG_EH;
656
				nr_timedout++;
657
			}
658 659 660 661 662 663 664
		} else {
			/* Normal completion occurred after
			 * SCSI timeout but before this point.
			 * Successfully complete it.
			 */
			scmd->retries = scmd->allowed;
			scsi_eh_finish_cmd(scmd, &ap->eh_done_q);
665
		}
666
	}
667

668 669 670 671 672
	/*
	 * If we have timed out qcs.  They belong to EH from
	 * this point but the state of the controller is
	 * unknown.  Freeze the port to make sure the IRQ
	 * handler doesn't diddle with those qcs.  This must
673
	 * be done atomically w.r.t. setting ATA_QCFLAG_EH.
674 675 676
	 */
	if (nr_timedout)
		__ata_port_freeze(ap);
677

678 679
	/* initialize eh_tries */
	ap->eh_tries = ATA_EH_MAX_TRIES;
680

681
	spin_unlock_irqrestore(ap->lock, flags);
682 683 684 685 686 687 688 689 690 691 692 693 694 695
}
EXPORT_SYMBOL(ata_scsi_cmd_error_handler);

/**
 * ata_scsi_port_error_handler - recover the port after the commands
 * @host:	SCSI host containing the port
 * @ap:		the ATA port
 *
 * Handle the recovery of the port @ap after all the commands
 * have been recovered.
 */
void ata_scsi_port_error_handler(struct Scsi_Host *host, struct ata_port *ap)
{
	unsigned long flags;
696
	struct ata_link *link;
697

698 699
	/* acquire EH ownership */
	ata_eh_acquire(ap);
700
 repeat:
701 702
	/* kill fast drain timer */
	del_timer_sync(&ap->fastdrain_timer);
Tejun Heo's avatar
Tejun Heo committed
703

704 705
	/* process port resume request */
	ata_eh_handle_port_resume(ap);
706

707 708
	/* fetch & clear EH info */
	spin_lock_irqsave(ap->lock, flags);
709

710 711 712
	ata_for_each_link(link, ap, HOST_FIRST) {
		struct ata_eh_context *ehc = &link->eh_context;
		struct ata_device *dev;
713

714 715 716
		memset(&link->eh_context, 0, sizeof(link->eh_context));
		link->eh_context.i = link->eh_info;
		memset(&link->eh_info, 0, sizeof(link->eh_info));
717

718 719
		ata_for_each_dev(dev, link, ENABLED) {
			int devno = dev->devno;
720

721 722 723
			ehc->saved_xfer_mode[devno] = dev->xfer_mode;
			if (ata_ncq_enabled(dev))
				ehc->saved_ncq_enabled |= 1 << devno;
724 725

			/* If we are resuming, wake up the device */
726 727
			if (ap->pflags & ATA_PFLAG_RESUMING) {
				dev->flags |= ATA_DFLAG_RESUMING;
728
				ehc->i.dev_action[devno] |= ATA_EH_SET_ACTIVE;
729
			}
730
		}
731
	}
732

733 734 735
	ap->pflags |= ATA_PFLAG_EH_IN_PROGRESS;
	ap->pflags &= ~ATA_PFLAG_EH_PENDING;
	ap->excl_link = NULL;	/* don't maintain exclusion over EH */
736

737
	spin_unlock_irqrestore(ap->lock, flags);
738

739 740 741 742 743 744 745 746 747 748
	/* invoke EH, skip if unloading or suspended */
	if (!(ap->pflags & (ATA_PFLAG_UNLOADING | ATA_PFLAG_SUSPENDED)))
		ap->ops->error_handler(ap);
	else {
		/* if unloading, commence suicide */
		if ((ap->pflags & ATA_PFLAG_UNLOADING) &&
		    !(ap->pflags & ATA_PFLAG_UNLOADED))
			ata_eh_unload(ap);
		ata_eh_finish(ap);
	}
749

750 751
	/* process port suspend request */
	ata_eh_handle_port_suspend(ap);
752

753 754 755 756 757
	/*
	 * Exception might have happened after ->error_handler recovered the
	 * port but before this point.  Repeat EH in such case.
	 */
	spin_lock_irqsave(ap->lock, flags);
758

759 760 761 762
	if (ap->pflags & ATA_PFLAG_EH_PENDING) {
		if (--ap->eh_tries) {
			spin_unlock_irqrestore(ap->lock, flags);
			goto repeat;
763
		}
764 765 766 767 768
		ata_port_err(ap,
			     "EH pending after %d tries, giving up\n",
			     ATA_EH_MAX_TRIES);
		ap->pflags &= ~ATA_PFLAG_EH_PENDING;
	}
769

770 771 772
	/* this run is complete, make sure EH info is clear */
	ata_for_each_link(link, ap, HOST_FIRST)
		memset(&link->eh_info, 0, sizeof(link->eh_info));
773

774 775 776 777 778 779
	/*
	 * end eh (clear host_eh_scheduled) while holding ap->lock such that if
	 * exception occurs after this point but before EH completion, SCSI
	 * midlayer will re-initiate EH.
	 */
	ap->ops->end_eh(ap);
780

781 782
	spin_unlock_irqrestore(ap->lock, flags);
	ata_eh_release(ap);
783 784 785

	scsi_eh_flush_done_q(&ap->eh_done_q);

786
	/* clean up */
787
	spin_lock_irqsave(ap->lock, flags);
788

789 790
	ap->pflags &= ~ATA_PFLAG_RESUMING;

791
	if (ap->pflags & ATA_PFLAG_LOADING)
792
		ap->pflags &= ~ATA_PFLAG_LOADING;
793 794
	else if ((ap->pflags & ATA_PFLAG_SCSI_HOTPLUG) &&
		!(ap->flags & ATA_FLAG_SAS_HOST))
795
		schedule_delayed_work(&ap->hotplug_task, 0);
796 797

	if (ap->pflags & ATA_PFLAG_RECOVERED)
798
		ata_port_info(ap, "EH complete\n");
799

800
	ap->pflags &= ~(ATA_PFLAG_SCSI_HOTPLUG | ATA_PFLAG_RECOVERED);
801

802
	/* tell wait_eh that we're done */
803
	ap->pflags &= ~ATA_PFLAG_EH_IN_PROGRESS;
804 805
	wake_up_all(&ap->eh_wait_q);

806
	spin_unlock_irqrestore(ap->lock, flags);
807
}
808
EXPORT_SYMBOL_GPL(ata_scsi_port_error_handler);
809

810 811 812 813 814 815 816 817 818 819 820 821 822 823 824
/**
 *	ata_port_wait_eh - Wait for the currently pending EH to complete
 *	@ap: Port to wait EH for
 *
 *	Wait until the currently pending EH is complete.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 */
void ata_port_wait_eh(struct ata_port *ap)
{
	unsigned long flags;
	DEFINE_WAIT(wait);

 retry:
825
	spin_lock_irqsave(ap->lock, flags);
826

827
	while (ap->pflags & (ATA_PFLAG_EH_PENDING | ATA_PFLAG_EH_IN_PROGRESS)) {
828
		prepare_to_wait(&ap->eh_wait_q, &wait, TASK_UNINTERRUPTIBLE);
829
		spin_unlock_irqrestore(ap->lock, flags);
830
		schedule();
831
		spin_lock_irqsave(ap->lock, flags);
832
	}
833
	finish_wait(&ap->eh_wait_q, &wait);
834

835
	spin_unlock_irqrestore(ap->lock, flags);
836 837

	/* make sure SCSI EH is complete */
Jeff Garzik's avatar
Jeff Garzik committed
838
	if (scsi_host_in_recovery(ap->scsi_host)) {
839
		ata_msleep(ap, 10);
840 841 842
		goto retry;
	}
}
843
EXPORT_SYMBOL_GPL(ata_port_wait_eh);
844

845
static unsigned int ata_eh_nr_in_flight(struct ata_port *ap)
Tejun Heo's avatar
Tejun Heo committed
846
{
847
	struct ata_queued_cmd *qc;
Tejun Heo's avatar
Tejun Heo committed
848
	unsigned int tag;
849
	unsigned int nr = 0;
Tejun Heo's avatar
Tejun Heo committed
850 851

	/* count only non-internal commands */
852 853
	ata_qc_for_each(ap, qc, tag) {
		if (qc)
Tejun Heo's avatar
Tejun Heo committed
854
			nr++;
855
	}
Tejun Heo's avatar
Tejun Heo committed
856 857 858 859

	return nr;
}

860
void ata_eh_fastdrain_timerfn(struct timer_list *t)
Tejun Heo's avatar
Tejun Heo committed
861
{
862
	struct ata_port *ap = from_timer(ap, t, fastdrain_timer);
Tejun Heo's avatar
Tejun Heo committed
863
	unsigned long flags;
864
	unsigned int cnt;
Tejun Heo's avatar
Tejun Heo committed
865 866 867 868 869 870 871 872 873 874

	spin_lock_irqsave(ap->lock, flags);

	cnt = ata_eh_nr_in_flight(ap);

	/* are we done? */
	if (!cnt)
		goto out_unlock;

	if (cnt == ap->fastdrain_cnt) {
875
		struct ata_queued_cmd *qc;
Tejun Heo's avatar
Tejun Heo committed
876 877 878 879 880
		unsigned int tag;

		/* No progress during the last interval, tag all
		 * in-flight qcs as timed out and freeze the port.
		 */
881
		ata_qc_for_each(ap, qc, tag) {
Tejun Heo's avatar
Tejun Heo committed
882 883 884 885 886 887 888 889 890
			if (qc)
				qc->err_mask |= AC_ERR_TIMEOUT;
		}

		ata_port_freeze(ap);
	} else {
		/* some qcs have finished, give it another chance */
		ap->fastdrain_cnt = cnt;
		ap->fastdrain_timer.expires =
891
			ata_deadline(jiffies, ATA_EH_FASTDRAIN_INTERVAL);
Tejun Heo's avatar
Tejun Heo committed
892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912
		add_timer(&ap->fastdrain_timer);
	}

 out_unlock:
	spin_unlock_irqrestore(ap->lock, flags);
}

/**
 *	ata_eh_set_pending - set ATA_PFLAG_EH_PENDING and activate fast drain
 *	@ap: target ATA port
 *	@fastdrain: activate fast drain
 *
 *	Set ATA_PFLAG_EH_PENDING and activate fast drain if @fastdrain
 *	is non-zero and EH wasn't pending before.  Fast drain ensures
 *	that EH kicks in in timely manner.
 *
 *	LOCKING:
 *	spin_lock_irqsave(host lock)
 */
static void ata_eh_set_pending(struct ata_port *ap, int fastdrain)
{
913
	unsigned int cnt;
Tejun Heo's avatar
Tejun Heo committed
914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930

	/* already scheduled? */
	if (ap->pflags & ATA_PFLAG_EH_PENDING)
		return;

	ap->pflags |= ATA_PFLAG_EH_PENDING;

	if (!fastdrain)
		return;

	/* do we have in-flight qcs? */
	cnt = ata_eh_nr_in_flight(ap);
	if (!cnt)
		return;

	/* activate fast drain */
	ap->fastdrain_cnt = cnt;
931 932
	ap->fastdrain_timer.expires =
		ata_deadline(jiffies, ATA_EH_FASTDRAIN_INTERVAL);
Tejun Heo's avatar
Tejun Heo committed
933 934 935
	add_timer(&ap->fastdrain_timer);
}

936 937 938 939 940 941 942 943
/**
 *	ata_qc_schedule_eh - schedule qc for error handling
 *	@qc: command to schedule error handling for
 *
 *	Schedule error handling for @qc.  EH will kick in as soon as
 *	other commands are drained.
 *
 *	LOCKING:
Jeff Garzik's avatar
Jeff Garzik committed
944
 *	spin_lock_irqsave(host lock)
945 946 947 948 949
 */
void ata_qc_schedule_eh(struct ata_queued_cmd *qc)
{
	struct ata_port *ap = qc->ap;

950
	qc->flags |= ATA_QCFLAG_EH;
Tejun Heo's avatar
Tejun Heo committed
951
	ata_eh_set_pending(ap, 1);
952 953 954

	/* The following will fail if timeout has already expired.
	 * ata_scsi_error() takes care of such scmds on EH entry.
955
	 * Note that ATA_QCFLAG_EH is unconditionally set after
956 957
	 * this function completes.
	 */
958
	blk_abort_request(scsi_cmd_to_rq(qc->scsicmd));
959 960
}

961
/**
962 963
 * ata_std_sched_eh - non-libsas ata_ports issue eh with this common routine
 * @ap: ATA port to schedule EH for
964
 *
965
 *	LOCKING: inherited from ata_port_schedule_eh
Jeff Garzik's avatar
Jeff Garzik committed
966
 *	spin_lock_irqsave(host lock)
967
 */
968
void ata_std_sched_eh(struct ata_port *ap)
969
{
970 971 972
	if (ap->pflags & ATA_PFLAG_INITIALIZING)
		return;

Tejun Heo's avatar
Tejun Heo committed
973
	ata_eh_set_pending(ap, 1);
Jeff Garzik's avatar
Jeff Garzik committed
974
	scsi_schedule_eh(ap->scsi_host);
975

976
	trace_ata_std_sched_eh(ap);
977
}
978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015
EXPORT_SYMBOL_GPL(ata_std_sched_eh);

/**
 * ata_std_end_eh - non-libsas ata_ports complete eh with this common routine
 * @ap: ATA port to end EH for
 *
 * In the libata object model there is a 1:1 mapping of ata_port to
 * shost, so host fields can be directly manipulated under ap->lock, in
 * the libsas case we need to hold a lock at the ha->level to coordinate
 * these events.
 *
 *	LOCKING:
 *	spin_lock_irqsave(host lock)
 */
void ata_std_end_eh(struct ata_port *ap)
{
	struct Scsi_Host *host = ap->scsi_host;

	host->host_eh_scheduled = 0;
}
EXPORT_SYMBOL(ata_std_end_eh);


/**
 *	ata_port_schedule_eh - schedule error handling without a qc
 *	@ap: ATA port to schedule EH for
 *
 *	Schedule error handling for @ap.  EH will kick in as soon as
 *	all commands are drained.
 *
 *	LOCKING:
 *	spin_lock_irqsave(host lock)
 */
void ata_port_schedule_eh(struct ata_port *ap)
{
	/* see: ata_std_sched_eh, unless you know better */
	ap->ops->sched_eh(ap);
}
1016
EXPORT_SYMBOL_GPL(ata_port_schedule_eh);
1017

1018
static int ata_do_link_abort(struct ata_port *ap, struct ata_link *link)
1019
{
1020
	struct ata_queued_cmd *qc;
1021 1022
	int tag, nr_aborted = 0;

Tejun Heo's avatar
Tejun Heo committed
1023 1024 1025
	/* we're gonna abort all commands, no need for fast drain */
	ata_eh_set_pending(ap, 0);

1026
	/* include internal tag in iteration */
1027
	ata_qc_for_each_with_internal(ap, qc, tag) {
1028
		if (qc && (!link || qc->dev->link == link)) {
1029
			qc->flags |= ATA_QCFLAG_EH;
1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040
			ata_qc_complete(qc);
			nr_aborted++;
		}
	}

	if (!nr_aborted)
		ata_port_schedule_eh(ap);

	return nr_aborted;
}

1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056
/**
 *	ata_link_abort - abort all qc's on the link
 *	@link: ATA link to abort qc's for
 *
 *	Abort all active qc's active on @link and schedule EH.
 *
 *	LOCKING:
 *	spin_lock_irqsave(host lock)
 *
 *	RETURNS:
 *	Number of aborted qc's.
 */
int ata_link_abort(struct ata_link *link)
{
	return ata_do_link_abort(link->ap, link);
}
1057
EXPORT_SYMBOL_GPL(ata_link_abort);
1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074

/**
 *	ata_port_abort - abort all qc's on the port
 *	@ap: ATA port to abort qc's for
 *
 *	Abort all active qc's of @ap and schedule EH.
 *
 *	LOCKING:
 *	spin_lock_irqsave(host_set lock)
 *
 *	RETURNS:
 *	Number of aborted qc's.
 */
int ata_port_abort(struct ata_port *ap)
{
	return ata_do_link_abort(ap, NULL);
}
1075
EXPORT_SYMBOL_GPL(ata_port_abort);
1076

1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092
/**
 *	__ata_port_freeze - freeze port
 *	@ap: ATA port to freeze
 *
 *	This function is called when HSM violation or some other
 *	condition disrupts normal operation of the port.  Frozen port
 *	is not allowed to perform any operation until the port is
 *	thawed, which usually follows a successful reset.
 *
 *	ap->ops->freeze() callback can be used for freezing the port
 *	hardware-wise (e.g. mask interrupt and stop DMA engine).  If a
 *	port cannot be frozen hardware-wise, the interrupt handler
 *	must ack and clear interrupts unconditionally while the port
 *	is frozen.
 *
 *	LOCKING:
Jeff Garzik's avatar
Jeff Garzik committed
1093
 *	spin_lock_irqsave(host lock)
1094 1095 1096 1097 1098 1099
 */
static void __ata_port_freeze(struct ata_port *ap)
{
	if (ap->ops->freeze)
		ap->ops->freeze(ap);

1100
	ap->pflags |= ATA_PFLAG_FROZEN;
1101

1102
	trace_ata_port_freeze(ap);
1103 1104 1105 1106 1107 1108
}

/**
 *	ata_port_freeze - abort & freeze port
 *	@ap: ATA port to freeze
 *
1109 1110 1111
 *	Abort and freeze @ap.  The freeze operation must be called
 *	first, because some hardware requires special operations
 *	before the taskfile registers are accessible.
1112 1113
 *
 *	LOCKING:
Jeff Garzik's avatar
Jeff Garzik committed
1114
 *	spin_lock_irqsave(host lock)
1115 1116 1117 1118 1119 1120 1121 1122
 *
 *	RETURNS:
 *	Number of aborted commands.
 */
int ata_port_freeze(struct ata_port *ap)
{
	__ata_port_freeze(ap);

1123
	return ata_port_abort(ap);
1124
}
1125
EXPORT_SYMBOL_GPL(ata_port_freeze);
1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139

/**
 *	ata_eh_freeze_port - EH helper to freeze port
 *	@ap: ATA port to freeze
 *
 *	Freeze @ap.
 *
 *	LOCKING:
 *	None.
 */
void ata_eh_freeze_port(struct ata_port *ap)
{
	unsigned long flags;

1140
	spin_lock_irqsave(ap->lock, flags);
1141
	__ata_port_freeze(ap);
1142
	spin_unlock_irqrestore(ap->lock, flags);
1143
}
1144
EXPORT_SYMBOL_GPL(ata_eh_freeze_port);
1145 1146

/**
1147
 *	ata_eh_thaw_port - EH helper to thaw port
1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158
 *	@ap: ATA port to thaw
 *
 *	Thaw frozen port @ap.
 *
 *	LOCKING:
 *	None.
 */
void ata_eh_thaw_port(struct ata_port *ap)
{
	unsigned long flags;

1159
	spin_lock_irqsave(ap->lock, flags);
1160

1161
	ap->pflags &= ~ATA_PFLAG_FROZEN;
1162 1163 1164 1165

	if (ap->ops->thaw)
		ap->ops->thaw(ap);

1166
	spin_unlock_irqrestore(ap->lock, flags);
1167

1168
	trace_ata_port_thaw(ap);
1169 1170
}

1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181
static void ata_eh_scsidone(struct scsi_cmnd *scmd)
{
	/* nada */
}

static void __ata_eh_qc_complete(struct ata_queued_cmd *qc)
{
	struct ata_port *ap = qc->ap;
	struct scsi_cmnd *scmd = qc->scsicmd;
	unsigned long flags;

1182
	spin_lock_irqsave(ap->lock, flags);
1183 1184 1185
	qc->scsidone = ata_eh_scsidone;
	__ata_qc_complete(qc);
	WARN_ON(ata_tag_valid(qc->tag));
1186
	spin_unlock_irqrestore(ap->lock, flags);
1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212

	scsi_eh_finish_cmd(scmd, &ap->eh_done_q);
}

/**
 *	ata_eh_qc_complete - Complete an active ATA command from EH
 *	@qc: Command to complete
 *
 *	Indicate to the mid and upper layers that an ATA command has
 *	completed.  To be used from EH.
 */
void ata_eh_qc_complete(struct ata_queued_cmd *qc)
{
	struct scsi_cmnd *scmd = qc->scsicmd;
	scmd->retries = scmd->allowed;
	__ata_eh_qc_complete(qc);
}

/**
 *	ata_eh_qc_retry - Tell midlayer to retry an ATA command after EH
 *	@qc: Command to retry
 *
 *	Indicate to the mid and upper layers that an ATA command
 *	should be retried.  To be used from EH.
 *
 *	SCSI midlayer limits the number of retries to scmd->allowed.
1213
 *	scmd->allowed is incremented for commands which get retried
1214 1215 1216 1217 1218
 *	due to unrelated failures (qc->err_mask is zero).
 */
void ata_eh_qc_retry(struct ata_queued_cmd *qc)
{
	struct scsi_cmnd *scmd = qc->scsicmd;
1219 1220
	if (!qc->err_mask)
		scmd->allowed++;
1221 1222
	__ata_eh_qc_complete(qc);
}
1223

1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237
/**
 *	ata_dev_disable - disable ATA device
 *	@dev: ATA device to disable
 *
 *	Disable @dev.
 *
 *	Locking:
 *	EH context.
 */
void ata_dev_disable(struct ata_device *dev)
{
	if (!ata_dev_enabled(dev))
		return;

1238
	ata_dev_warn(dev, "disable device\n");
1239

1240
	ata_eh_dev_disable(dev);
1241
}
1242
EXPORT_SYMBOL_GPL(ata_dev_disable);
1243

1244 1245 1246 1247 1248 1249 1250 1251 1252
/**
 *	ata_eh_detach_dev - detach ATA device
 *	@dev: ATA device to detach
 *
 *	Detach @dev.
 *
 *	LOCKING:
 *	None.
 */
1253
void ata_eh_detach_dev(struct ata_device *dev)
1254
{
1255 1256
	struct ata_link *link = dev->link;
	struct ata_port *ap = link->ap;
1257
	struct ata_eh_context *ehc = &link->eh_context;
1258 1259
	unsigned long flags;

1260 1261
	/*
	 * If the device is still enabled, transition it to standby power mode
1262
	 * (i.e. spin down HDDs) and disable it.
1263
	 */
1264
	if (ata_dev_enabled(dev)) {
1265
		ata_dev_power_set_standby(dev);
1266 1267
		ata_eh_dev_disable(dev);
	}
1268

1269
	spin_lock_irqsave(ap->lock, flags);
1270 1271 1272 1273 1274

	dev->flags &= ~ATA_DFLAG_DETACH;

	if (ata_scsi_offline_dev(dev)) {
		dev->flags |= ATA_DFLAG_DETACHED;
1275
		ap->pflags |= ATA_PFLAG_SCSI_HOTPLUG;
1276 1277
	}

1278
	/* clear per-dev EH info */
1279 1280
	ata_eh_clear_action(link, dev, &link->eh_info, ATA_EH_PERDEV_MASK);
	ata_eh_clear_action(link, dev, &link->eh_context.i, ATA_EH_PERDEV_MASK);
1281 1282
	ehc->saved_xfer_mode[dev->devno] = 0;
	ehc->saved_ncq_enabled &= ~(1 << dev->devno);
1283

1284
	spin_unlock_irqrestore(ap->lock, flags);
1285 1286
}

1287 1288
/**
 *	ata_eh_about_to_do - about to perform eh_action
1289
 *	@link: target ATA link
1290
 *	@dev: target ATA dev for per-dev action (can be NULL)
1291 1292 1293
 *	@action: action about to be performed
 *
 *	Called just before performing EH actions to clear related bits
1294 1295
 *	in @link->eh_info such that eh actions are not unnecessarily
 *	repeated.
1296 1297 1298 1299
 *
 *	LOCKING:
 *	None.
 */
1300 1301
void ata_eh_about_to_do(struct ata_link *link, struct ata_device *dev,
			unsigned int action)
1302
{
1303 1304 1305
	struct ata_port *ap = link->ap;
	struct ata_eh_info *ehi = &link->eh_info;
	struct ata_eh_context *ehc = &link->eh_context;
1306 1307
	unsigned long flags;

1308 1309
	trace_ata_eh_about_to_do(link, dev ? dev->devno : 0, action);

1310
	spin_lock_irqsave(ap->lock, flags);
1311

1312
	ata_eh_clear_action(link, dev, ehi, action);
1313

1314 1315 1316 1317
	/* About to take EH action, set RECOVERED.  Ignore actions on
	 * slave links as master will do them again.
	 */
	if (!(ehc->i.flags & ATA_EHI_QUIET) && link != ap->slave_link)
1318 1319
		ap->pflags |= ATA_PFLAG_RECOVERED;

1320
	spin_unlock_irqrestore(ap->lock, flags);
1321 1322
}

1323 1324
/**
 *	ata_eh_done - EH action complete
1325
 *	@link: ATA link for which EH actions are complete
1326 1327 1328 1329
 *	@dev: target ATA dev for per-dev action (can be NULL)
 *	@action: action just completed
 *
 *	Called right after performing EH actions to clear related bits
1330
 *	in @link->eh_context.
1331 1332 1333 1334
 *
 *	LOCKING:
 *	None.
 */
1335 1336
void ata_eh_done(struct ata_link *link, struct ata_device *dev,
		 unsigned int action)
1337
{
1338
	struct ata_eh_context *ehc = &link->eh_context;
Tejun Heo's avatar
Tejun Heo committed
1339

1340 1341
	trace_ata_eh_done(link, dev ? dev->devno : 0, action);

1342
	ata_eh_clear_action(link, dev, &ehc->i, action);
1343 1344
}

1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358
/**
 *	ata_err_string - convert err_mask to descriptive string
 *	@err_mask: error mask to convert to string
 *
 *	Convert @err_mask to descriptive string.  Errors are
 *	prioritized according to severity and only the most severe
 *	error is reported.
 *
 *	LOCKING:
 *	None.
 *
 *	RETURNS:
 *	Descriptive string for @err_mask
 */
1359
static const char *ata_err_string(unsigned int err_mask)
1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376
{
	if (err_mask & AC_ERR_HOST_BUS)
		return "host bus error";
	if (err_mask & AC_ERR_ATA_BUS)
		return "ATA bus error";
	if (err_mask & AC_ERR_TIMEOUT)
		return "timeout";
	if (err_mask & AC_ERR_HSM)
		return "HSM violation";
	if (err_mask & AC_ERR_SYSTEM)
		return "internal error";
	if (err_mask & AC_ERR_MEDIA)
		return "media error";
	if (err_mask & AC_ERR_INVALID)
		return "invalid argument";
	if (err_mask & AC_ERR_DEV)
		return "device error";
1377 1378 1379 1380
	if (err_mask & AC_ERR_NCQ)
		return "NCQ error";
	if (err_mask & AC_ERR_NODEV_HINT)
		return "Polling detection error";
1381 1382 1383
	return "unknown error";
}

1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396
/**
 *	atapi_eh_tur - perform ATAPI TEST_UNIT_READY
 *	@dev: target ATAPI device
 *	@r_sense_key: out parameter for sense_key
 *
 *	Perform ATAPI TEST_UNIT_READY.
 *
 *	LOCKING:
 *	EH context (may sleep).
 *
 *	RETURNS:
 *	0 on success, AC_ERR_* mask on failure.
 */
1397
unsigned int atapi_eh_tur(struct ata_device *dev, u8 *r_sense_key)
1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410
{
	u8 cdb[ATAPI_CDB_LEN] = { TEST_UNIT_READY, 0, 0, 0, 0, 0 };
	struct ata_taskfile tf;
	unsigned int err_mask;

	ata_tf_init(dev, &tf);

	tf.flags |= ATA_TFLAG_ISADDR | ATA_TFLAG_DEVICE;
	tf.command = ATA_CMD_PACKET;
	tf.protocol = ATAPI_PROT_NODATA;

	err_mask = ata_exec_internal(dev, &tf, cdb, DMA_NONE, NULL, 0, 0);
	if (err_mask == AC_ERR_DEV)
1411
		*r_sense_key = tf.error >> 4;
1412 1413 1414
	return err_mask;
}

1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451
/**
 *	ata_eh_decide_disposition - Disposition a qc based on sense data
 *	@qc: qc to examine
 *
 *	For a regular SCSI command, the SCSI completion callback (scsi_done())
 *	will call scsi_complete(), which will call scsi_decide_disposition(),
 *	which will call scsi_check_sense(). scsi_complete() finally calls
 *	scsi_finish_command(). This is fine for SCSI, since any eventual sense
 *	data is usually returned in the completion itself (without invoking SCSI
 *	EH). However, for a QC, we always need to fetch the sense data
 *	explicitly using SCSI EH.
 *
 *	A command that is completed via SCSI EH will instead be completed using
 *	scsi_eh_flush_done_q(), which will call scsi_finish_command() directly
 *	(without ever calling scsi_check_sense()).
 *
 *	For a command that went through SCSI EH, it is the responsibility of the
 *	SCSI EH strategy handler to call scsi_decide_disposition(), see e.g. how
 *	scsi_eh_get_sense() calls scsi_decide_disposition() for SCSI LLDDs that
 *	do not get the sense data as part of the completion.
 *
 *	Thus, for QC commands that went via SCSI EH, we need to call
 *	scsi_check_sense() ourselves, similar to how scsi_eh_get_sense() calls
 *	scsi_decide_disposition(), which calls scsi_check_sense(), in order to
 *	set the correct SCSI ML byte (if any).
 *
 *	LOCKING:
 *	EH context.
 *
 *	RETURNS:
 *	SUCCESS or FAILED or NEEDS_RETRY or ADD_TO_MLQUEUE
 */
enum scsi_disposition ata_eh_decide_disposition(struct ata_queued_cmd *qc)
{
	return scsi_check_sense(qc->scsicmd);
}

1452 1453
/**
 *	ata_eh_request_sense - perform REQUEST_SENSE_DATA_EXT
1454
 *	@qc: qc to perform REQUEST_SENSE_SENSE_DATA_EXT to
1455 1456 1457 1458 1459 1460
 *
 *	Perform REQUEST_SENSE_DATA_EXT after the device reported CHECK
 *	SENSE.  This function is an EH helper.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
1461 1462 1463
 *
 *	RETURNS:
 *	true if sense data could be fetched, false otherwise.
1464
 */
1465
static bool ata_eh_request_sense(struct ata_queued_cmd *qc)
1466
{
1467
	struct scsi_cmnd *cmd = qc->scsicmd;
1468 1469 1470 1471
	struct ata_device *dev = qc->dev;
	struct ata_taskfile tf;
	unsigned int err_mask;

1472
	if (ata_port_is_frozen(qc->ap)) {
1473
		ata_dev_warn(dev, "sense data available but port frozen\n");
1474
		return false;
1475 1476 1477 1478
	}

	if (!ata_id_sense_reporting_enabled(dev->id)) {
		ata_dev_warn(qc->dev, "sense data reporting disabled\n");
1479
		return false;
1480 1481 1482 1483 1484 1485 1486 1487 1488 1489
	}

	ata_tf_init(dev, &tf);
	tf.flags |= ATA_TFLAG_ISADDR | ATA_TFLAG_DEVICE;
	tf.flags |= ATA_TFLAG_LBA | ATA_TFLAG_LBA48;
	tf.command = ATA_CMD_REQ_SENSE_DATA;
	tf.protocol = ATA_PROT_NODATA;

	err_mask = ata_exec_internal(dev, &tf, NULL, DMA_NONE, NULL, 0, 0);
	/* Ignore err_mask; ATA_ERR might be set */
1490
	if (tf.status & ATA_SENSE) {
1491
		if (ata_scsi_sense_is_valid(tf.lbah, tf.lbam, tf.lbal)) {
1492 1493 1494 1495
			/* Set sense without also setting scsicmd->result */
			scsi_build_sense_buffer(dev->flags & ATA_DFLAG_D_SENSE,
						cmd->sense_buffer, tf.lbah,
						tf.lbam, tf.lbal);
1496
			qc->flags |= ATA_QCFLAG_SENSE_VALID;
1497
			return true;
1498
		}
1499 1500
	} else {
		ata_dev_warn(dev, "request sense failed stat %02x emask %x\n",
1501
			     tf.status, err_mask);
1502
	}
1503 1504

	return false;
1505 1506
}

1507 1508 1509 1510
/**
 *	atapi_eh_request_sense - perform ATAPI REQUEST_SENSE
 *	@dev: device to perform REQUEST_SENSE to
 *	@sense_buf: result sense data buffer (SCSI_SENSE_BUFFERSIZE bytes long)
1511
 *	@dfl_sense_key: default sense key to use
1512 1513 1514 1515 1516 1517 1518 1519 1520 1521
 *
 *	Perform ATAPI REQUEST_SENSE after the device reported CHECK
 *	SENSE.  This function is EH helper.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 *
 *	RETURNS:
 *	0 on success, AC_ERR_* mask on failure
 */
1522
unsigned int atapi_eh_request_sense(struct ata_device *dev,
1523
					   u8 *sense_buf, u8 dfl_sense_key)
1524
{
1525 1526
	u8 cdb[ATAPI_CDB_LEN] =
		{ REQUEST_SENSE, 0, 0, 0, SCSI_SENSE_BUFFERSIZE, 0 };
Tejun Heo's avatar
Tejun Heo committed
1527
	struct ata_port *ap = dev->link->ap;
1528 1529 1530 1531
	struct ata_taskfile tf;

	memset(sense_buf, 0, SCSI_SENSE_BUFFERSIZE);

1532 1533 1534
	/* initialize sense_buf with the error register,
	 * for the case where they are -not- overwritten
	 */
1535
	sense_buf[0] = 0x70;
1536
	sense_buf[2] = dfl_sense_key;
1537

1538
	/* some devices time out if garbage left in tf */
1539
	ata_tf_init(dev, &tf);
1540 1541 1542 1543 1544 1545

	tf.flags |= ATA_TFLAG_ISADDR | ATA_TFLAG_DEVICE;
	tf.command = ATA_CMD_PACKET;

	/* is it pointless to prefer PIO for "safety reasons"? */
	if (ap->flags & ATA_FLAG_PIO_DMA) {
1546
		tf.protocol = ATAPI_PROT_DMA;
1547 1548
		tf.feature |= ATAPI_PKT_DMA;
	} else {
1549
		tf.protocol = ATAPI_PROT_PIO;
1550 1551
		tf.lbam = SCSI_SENSE_BUFFERSIZE;
		tf.lbah = 0;
1552 1553 1554
	}

	return ata_exec_internal(dev, &tf, cdb, DMA_FROM_DEVICE,
1555
				 sense_buf, SCSI_SENSE_BUFFERSIZE, 0);
1556 1557 1558 1559
}

/**
 *	ata_eh_analyze_serror - analyze SError for a failed port
1560
 *	@link: ATA link to analyze SError for
1561 1562 1563 1564 1565 1566 1567
 *
 *	Analyze SError if available and further determine cause of
 *	failure.
 *
 *	LOCKING:
 *	None.
 */
1568
static void ata_eh_analyze_serror(struct ata_link *link)
1569
{
1570
	struct ata_eh_context *ehc = &link->eh_context;
1571 1572
	u32 serror = ehc->i.serror;
	unsigned int err_mask = 0, action = 0;
1573
	u32 hotplug_mask;
1574

1575
	if (serror & (SERR_PERSISTENT | SERR_DATA)) {
1576
		err_mask |= AC_ERR_ATA_BUS;
Tejun Heo's avatar
Tejun Heo committed
1577
		action |= ATA_EH_RESET;
1578 1579 1580
	}
	if (serror & SERR_PROTOCOL) {
		err_mask |= AC_ERR_HSM;
Tejun Heo's avatar
Tejun Heo committed
1581
		action |= ATA_EH_RESET;
1582 1583 1584
	}
	if (serror & SERR_INTERNAL) {
		err_mask |= AC_ERR_SYSTEM;
Tejun Heo's avatar
Tejun Heo committed
1585
		action |= ATA_EH_RESET;
1586
	}
1587 1588 1589 1590 1591 1592

	/* Determine whether a hotplug event has occurred.  Both
	 * SError.N/X are considered hotplug events for enabled or
	 * host links.  For disabled PMP links, only N bit is
	 * considered as X bit is left at 1 for link plugging.
	 */
1593
	if (link->lpm_policy > ATA_LPM_MAX_POWER)
1594 1595
		hotplug_mask = 0;	/* hotplug doesn't work w/ LPM */
	else if (!(link->flags & ATA_LFLAG_DISABLED) || ata_is_host_link(link))
1596 1597 1598 1599 1600
		hotplug_mask = SERR_PHYRDY_CHG | SERR_DEV_XCHG;
	else
		hotplug_mask = SERR_PHYRDY_CHG;

	if (serror & hotplug_mask)
1601
		ata_ehi_hotplugged(&ehc->i);
1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612

	ehc->i.err_mask |= err_mask;
	ehc->i.action |= action;
}

/**
 *	ata_eh_analyze_tf - analyze taskfile of a failed qc
 *	@qc: qc to analyze
 *
 *	Analyze taskfile of @qc and further determine cause of
 *	failure.  This function also requests ATAPI sense data if
Lucas De Marchi's avatar
Lucas De Marchi committed
1613
 *	available.
1614 1615 1616 1617 1618 1619 1620
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 *
 *	RETURNS:
 *	Determined recovery action
 */
1621
static unsigned int ata_eh_analyze_tf(struct ata_queued_cmd *qc)
1622
{
1623
	const struct ata_taskfile *tf = &qc->result_tf;
1624
	unsigned int tmp, action = 0;
1625
	u8 stat = tf->status, err = tf->error;
1626 1627 1628

	if ((stat & (ATA_BUSY | ATA_DRQ | ATA_DRDY)) != ATA_DRDY) {
		qc->err_mask |= AC_ERR_HSM;
Tejun Heo's avatar
Tejun Heo committed
1629
		return ATA_EH_RESET;
1630 1631
	}

1632
	if (stat & (ATA_ERR | ATA_DF)) {
1633
		qc->err_mask |= AC_ERR_DEV;
1634 1635 1636 1637 1638 1639 1640
		/*
		 * Sense data reporting does not work if the
		 * device fault bit is set.
		 */
		if (stat & ATA_DF)
			stat &= ~ATA_SENSE;
	} else {
1641
		return 0;
1642
	}
1643 1644

	switch (qc->dev->class) {
1645
	case ATA_DEV_ATA:
1646
	case ATA_DEV_ZAC:
1647 1648 1649 1650 1651 1652 1653
		/*
		 * Fetch the sense data explicitly if:
		 * -It was a non-NCQ command that failed, or
		 * -It was a NCQ command that failed, but the sense data
		 *  was not included in the NCQ command error log
		 *  (i.e. NCQ autosense is not supported by the device).
		 */
1654 1655 1656
		if (!(qc->flags & ATA_QCFLAG_SENSE_VALID) &&
		    (stat & ATA_SENSE) && ata_eh_request_sense(qc))
			set_status_byte(qc->scsicmd, SAM_STAT_CHECK_CONDITION);
1657 1658
		if (err & ATA_ICRC)
			qc->err_mask |= AC_ERR_ATA_BUS;
1659
		if (err & (ATA_UNC | ATA_AMNF))
1660 1661 1662 1663 1664 1665
			qc->err_mask |= AC_ERR_MEDIA;
		if (err & ATA_IDNF)
			qc->err_mask |= AC_ERR_INVALID;
		break;

	case ATA_DEV_ATAPI:
1666
		if (!ata_port_is_frozen(qc->ap)) {
1667 1668
			tmp = atapi_eh_request_sense(qc->dev,
						qc->scsicmd->sense_buffer,
1669
						qc->result_tf.error >> 4);
1670
			if (!tmp)
1671
				qc->flags |= ATA_QCFLAG_SENSE_VALID;
1672
			else
1673 1674
				qc->err_mask |= tmp;
		}
1675 1676
	}

1677
	if (qc->flags & ATA_QCFLAG_SENSE_VALID) {
1678 1679
		enum scsi_disposition ret = ata_eh_decide_disposition(qc);

1680
		/*
1681
		 * SUCCESS here means that the sense code could be
1682 1683
		 * evaluated and should be passed to the upper layers
		 * for correct evaluation.
1684
		 * FAILED means the sense code could not be interpreted
1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695
		 * and the device would need to be reset.
		 * NEEDS_RETRY and ADD_TO_MLQUEUE means that the
		 * command would need to be retried.
		 */
		if (ret == NEEDS_RETRY || ret == ADD_TO_MLQUEUE) {
			qc->flags |= ATA_QCFLAG_RETRY;
			qc->err_mask |= AC_ERR_OTHER;
		} else if (ret != SUCCESS) {
			qc->err_mask |= AC_ERR_HSM;
		}
	}
1696
	if (qc->err_mask & (AC_ERR_HSM | AC_ERR_TIMEOUT | AC_ERR_ATA_BUS))
Tejun Heo's avatar
Tejun Heo committed
1697
		action |= ATA_EH_RESET;
1698 1699 1700 1701

	return action;
}

1702 1703
static int ata_eh_categorize_error(unsigned int eflags, unsigned int err_mask,
				   int *xfer_ok)
1704
{
1705 1706 1707 1708 1709 1710
	int base = 0;

	if (!(eflags & ATA_EFLAG_DUBIOUS_XFER))
		*xfer_ok = 1;

	if (!*xfer_ok)
1711
		base = ATA_ECAT_DUBIOUS_NONE;
1712

1713
	if (err_mask & AC_ERR_ATA_BUS)
1714
		return base + ATA_ECAT_ATA_BUS;
1715

1716
	if (err_mask & AC_ERR_TIMEOUT)
1717
		return base + ATA_ECAT_TOUT_HSM;
1718

1719
	if (eflags & ATA_EFLAG_IS_IO) {
1720
		if (err_mask & AC_ERR_HSM)
1721
			return base + ATA_ECAT_TOUT_HSM;
1722 1723
		if ((err_mask &
		     (AC_ERR_DEV|AC_ERR_MEDIA|AC_ERR_INVALID)) == AC_ERR_DEV)
1724
			return base + ATA_ECAT_UNK_DEV;
1725 1726 1727 1728 1729
	}

	return 0;
}

1730
struct speed_down_verdict_arg {
1731
	u64 since;
1732
	int xfer_ok;
1733
	int nr_errors[ATA_ECAT_NR];
1734 1735
};

1736
static int speed_down_verdict_cb(struct ata_ering_entry *ent, void *void_arg)
1737
{
1738
	struct speed_down_verdict_arg *arg = void_arg;
1739
	int cat;
1740

1741
	if ((ent->eflags & ATA_EFLAG_OLD_ER) || (ent->timestamp < arg->since))
1742 1743
		return -1;

1744 1745
	cat = ata_eh_categorize_error(ent->eflags, ent->err_mask,
				      &arg->xfer_ok);
1746
	arg->nr_errors[cat]++;
1747

1748 1749 1750 1751
	return 0;
}

/**
1752
 *	ata_eh_speed_down_verdict - Determine speed down verdict
1753 1754 1755
 *	@dev: Device of interest
 *
 *	This function examines error ring of @dev and determines
1756 1757
 *	whether NCQ needs to be turned off, transfer speed should be
 *	stepped down, or falling back to PIO is necessary.
1758
 *
1759 1760 1761 1762 1763 1764 1765
 *	ECAT_ATA_BUS	: ATA_BUS error for any command
 *
 *	ECAT_TOUT_HSM	: TIMEOUT for any command or HSM violation for
 *			  IO commands
 *
 *	ECAT_UNK_DEV	: Unknown DEV error for IO commands
 *
1766 1767 1768
 *	ECAT_DUBIOUS_*	: Identical to above three but occurred while
 *			  data transfer hasn't been verified.
 *
1769 1770 1771
 *	Verdicts are
 *
 *	NCQ_OFF		: Turn off NCQ.
1772
 *
1773 1774
 *	SPEED_DOWN	: Speed down transfer speed but don't fall back
 *			  to PIO.
1775
 *
1776
 *	FALLBACK_TO_PIO	: Fall back to PIO.
1777
 *
1778
 *	Even if multiple verdicts are returned, only one action is
1779 1780 1781 1782 1783
 *	taken per error.  An action triggered by non-DUBIOUS errors
 *	clears ering, while one triggered by DUBIOUS_* errors doesn't.
 *	This is to expedite speed down decisions right after device is
 *	initially configured.
 *
1784
 *	The following are speed down rules.  #1 and #2 deal with
1785
 *	DUBIOUS errors.
1786
 *
1787 1788 1789 1790 1791 1792 1793
 *	1. If more than one DUBIOUS_ATA_BUS or DUBIOUS_TOUT_HSM errors
 *	   occurred during last 5 mins, SPEED_DOWN and FALLBACK_TO_PIO.
 *
 *	2. If more than one DUBIOUS_TOUT_HSM or DUBIOUS_UNK_DEV errors
 *	   occurred during last 5 mins, NCQ_OFF.
 *
 *	3. If more than 8 ATA_BUS, TOUT_HSM or UNK_DEV errors
Lucas De Marchi's avatar
Lucas De Marchi committed
1794
 *	   occurred during last 5 mins, FALLBACK_TO_PIO
1795
 *
1796
 *	4. If more than 3 TOUT_HSM or UNK_DEV errors occurred
1797 1798
 *	   during last 10 mins, NCQ_OFF.
 *
1799
 *	5. If more than 3 ATA_BUS or TOUT_HSM errors, or more than 6
1800
 *	   UNK_DEV errors occurred during last 10 mins, SPEED_DOWN.
1801
 *
1802 1803 1804 1805
 *	LOCKING:
 *	Inherited from caller.
 *
 *	RETURNS:
1806
 *	OR of ATA_EH_SPDN_* flags.
1807
 */
1808
static unsigned int ata_eh_speed_down_verdict(struct ata_device *dev)
1809
{
1810 1811 1812 1813
	const u64 j5mins = 5LLU * 60 * HZ, j10mins = 10LLU * 60 * HZ;
	u64 j64 = get_jiffies_64();
	struct speed_down_verdict_arg arg;
	unsigned int verdict = 0;
1814

1815
	/* scan past 5 mins of error history */
1816
	memset(&arg, 0, sizeof(arg));
1817
	arg.since = j64 - min(j64, j5mins);
1818
	ata_ering_map(&dev->ering, speed_down_verdict_cb, &arg);
1819

1820 1821 1822 1823 1824 1825 1826 1827 1828
	if (arg.nr_errors[ATA_ECAT_DUBIOUS_ATA_BUS] +
	    arg.nr_errors[ATA_ECAT_DUBIOUS_TOUT_HSM] > 1)
		verdict |= ATA_EH_SPDN_SPEED_DOWN |
			ATA_EH_SPDN_FALLBACK_TO_PIO | ATA_EH_SPDN_KEEP_ERRORS;

	if (arg.nr_errors[ATA_ECAT_DUBIOUS_TOUT_HSM] +
	    arg.nr_errors[ATA_ECAT_DUBIOUS_UNK_DEV] > 1)
		verdict |= ATA_EH_SPDN_NCQ_OFF | ATA_EH_SPDN_KEEP_ERRORS;

1829 1830
	if (arg.nr_errors[ATA_ECAT_ATA_BUS] +
	    arg.nr_errors[ATA_ECAT_TOUT_HSM] +
Tejun Heo's avatar
Tejun Heo committed
1831
	    arg.nr_errors[ATA_ECAT_UNK_DEV] > 6)
1832
		verdict |= ATA_EH_SPDN_FALLBACK_TO_PIO;
1833

1834
	/* scan past 10 mins of error history */
1835
	memset(&arg, 0, sizeof(arg));
1836
	arg.since = j64 - min(j64, j10mins);
1837
	ata_ering_map(&dev->ering, speed_down_verdict_cb, &arg);
1838

1839 1840 1841 1842 1843 1844
	if (arg.nr_errors[ATA_ECAT_TOUT_HSM] +
	    arg.nr_errors[ATA_ECAT_UNK_DEV] > 3)
		verdict |= ATA_EH_SPDN_NCQ_OFF;

	if (arg.nr_errors[ATA_ECAT_ATA_BUS] +
	    arg.nr_errors[ATA_ECAT_TOUT_HSM] > 3 ||
Tejun Heo's avatar
Tejun Heo committed
1845
	    arg.nr_errors[ATA_ECAT_UNK_DEV] > 6)
1846
		verdict |= ATA_EH_SPDN_SPEED_DOWN;
1847

1848
	return verdict;
1849 1850 1851 1852 1853
}

/**
 *	ata_eh_speed_down - record error and speed down if necessary
 *	@dev: Failed device
1854
 *	@eflags: mask of ATA_EFLAG_* flags
1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865
 *	@err_mask: err_mask of the error
 *
 *	Record error and examine error history to determine whether
 *	adjusting transmission speed is necessary.  It also sets
 *	transmission limits appropriately if such adjustment is
 *	necessary.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 *
 *	RETURNS:
1866
 *	Determined recovery action.
1867
 */
1868 1869
static unsigned int ata_eh_speed_down(struct ata_device *dev,
				unsigned int eflags, unsigned int err_mask)
1870
{
Tejun Heo's avatar
Tejun Heo committed
1871
	struct ata_link *link = ata_dev_phys_link(dev);
1872
	int xfer_ok = 0;
1873 1874 1875 1876
	unsigned int verdict;
	unsigned int action = 0;

	/* don't bother if Cat-0 error */
1877
	if (ata_eh_categorize_error(eflags, err_mask, &xfer_ok) == 0)
1878 1879 1880
		return 0;

	/* record error and determine whether speed down is necessary */
1881
	ata_ering_record(&dev->ering, eflags, err_mask);
1882
	verdict = ata_eh_speed_down_verdict(dev);
1883

1884
	/* turn off NCQ? */
1885
	if ((verdict & ATA_EH_SPDN_NCQ_OFF) && ata_ncq_enabled(dev)) {
1886
		dev->flags |= ATA_DFLAG_NCQ_OFF;
1887
		ata_dev_warn(dev, "NCQ disabled due to excessive errors\n");
1888 1889
		goto done;
	}
1890

1891 1892 1893
	/* speed down? */
	if (verdict & ATA_EH_SPDN_SPEED_DOWN) {
		/* speed down SATA link speed if possible */
1894
		if (sata_down_spd_limit(link, 0) == 0) {
Tejun Heo's avatar
Tejun Heo committed
1895
			action |= ATA_EH_RESET;
1896 1897
			goto done;
		}
1898

1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914
		/* lower transfer mode */
		if (dev->spdn_cnt < 2) {
			static const int dma_dnxfer_sel[] =
				{ ATA_DNXFER_DMA, ATA_DNXFER_40C };
			static const int pio_dnxfer_sel[] =
				{ ATA_DNXFER_PIO, ATA_DNXFER_FORCE_PIO0 };
			int sel;

			if (dev->xfer_shift != ATA_SHIFT_PIO)
				sel = dma_dnxfer_sel[dev->spdn_cnt];
			else
				sel = pio_dnxfer_sel[dev->spdn_cnt];

			dev->spdn_cnt++;

			if (ata_down_xfermask_limit(dev, sel) == 0) {
Tejun Heo's avatar
Tejun Heo committed
1915
				action |= ATA_EH_RESET;
1916 1917 1918 1919 1920 1921
				goto done;
			}
		}
	}

	/* Fall back to PIO?  Slowing down to PIO is meaningless for
Tejun Heo's avatar
Tejun Heo committed
1922
	 * SATA ATA devices.  Consider it only for PATA and SATAPI.
1923 1924
	 */
	if ((verdict & ATA_EH_SPDN_FALLBACK_TO_PIO) && (dev->spdn_cnt >= 2) &&
Tejun Heo's avatar
Tejun Heo committed
1925
	    (link->ap->cbl != ATA_CBL_SATA || dev->class == ATA_DEV_ATAPI) &&
1926 1927 1928
	    (dev->xfer_shift != ATA_SHIFT_PIO)) {
		if (ata_down_xfermask_limit(dev, ATA_DNXFER_FORCE_PIO) == 0) {
			dev->spdn_cnt = 0;
Tejun Heo's avatar
Tejun Heo committed
1929
			action |= ATA_EH_RESET;
1930 1931 1932
			goto done;
		}
	}
1933 1934

	return 0;
1935 1936
 done:
	/* device has been slowed down, blow error history */
1937 1938
	if (!(verdict & ATA_EH_SPDN_KEEP_ERRORS))
		ata_ering_clear(&dev->ering);
1939
	return action;
1940 1941
}

1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952
/**
 *	ata_eh_worth_retry - analyze error and decide whether to retry
 *	@qc: qc to possibly retry
 *
 *	Look at the cause of the error and decide if a retry
 * 	might be useful or not.  We don't want to retry media errors
 *	because the drive itself has probably already taken 10-30 seconds
 *	doing its own internal retries before reporting the failure.
 */
static inline int ata_eh_worth_retry(struct ata_queued_cmd *qc)
{
1953
	if (qc->err_mask & AC_ERR_MEDIA)
1954 1955 1956 1957 1958 1959 1960 1961
		return 0;	/* don't retry media errors */
	if (qc->flags & ATA_QCFLAG_IO)
		return 1;	/* otherwise retry anything from fs stack */
	if (qc->err_mask & AC_ERR_INVALID)
		return 0;	/* don't retry these */
	return qc->err_mask != AC_ERR_DEV;  /* retry if not dev error */
}

1962 1963 1964 1965 1966 1967 1968 1969 1970
/**
 *      ata_eh_quiet - check if we need to be quiet about a command error
 *      @qc: qc to check
 *
 *      Look at the qc flags anbd its scsi command request flags to determine
 *      if we need to be quiet about the command failure.
 */
static inline bool ata_eh_quiet(struct ata_queued_cmd *qc)
{
1971
	if (qc->scsicmd && scsi_cmd_to_rq(qc->scsicmd)->rq_flags & RQF_QUIET)
1972 1973 1974 1975
		qc->flags |= ATA_QCFLAG_QUIET;
	return qc->flags & ATA_QCFLAG_QUIET;
}

1976
static int ata_eh_get_non_ncq_success_sense(struct ata_link *link)
1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993
{
	struct ata_port *ap = link->ap;
	struct ata_queued_cmd *qc;

	qc = __ata_qc_from_tag(ap, link->active_tag);
	if (!qc)
		return -EIO;

	if (!(qc->flags & ATA_QCFLAG_EH) ||
	    !(qc->flags & ATA_QCFLAG_EH_SUCCESS_CMD) ||
	    qc->err_mask)
		return -EIO;

	if (!ata_eh_request_sense(qc))
		return -EIO;

	/*
1994 1995
	 * No point in checking the return value, since the command has already
	 * completed successfully.
1996
	 */
1997
	ata_eh_decide_disposition(qc);
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026

	return 0;
}

static void ata_eh_get_success_sense(struct ata_link *link)
{
	struct ata_eh_context *ehc = &link->eh_context;
	struct ata_device *dev = link->device;
	struct ata_port *ap = link->ap;
	struct ata_queued_cmd *qc;
	int tag, ret = 0;

	if (!(ehc->i.dev_action[dev->devno] & ATA_EH_GET_SUCCESS_SENSE))
		return;

	/* if frozen, we can't do much */
	if (ata_port_is_frozen(ap)) {
		ata_dev_warn(dev,
			"successful sense data available but port frozen\n");
		goto out;
	}

	/*
	 * If the link has sactive set, then we have outstanding NCQ commands
	 * and have to read the Successful NCQ Commands log to get the sense
	 * data. Otherwise, we are dealing with a non-NCQ command and use
	 * request sense ext command to retrieve the sense data.
	 */
	if (link->sactive)
2027
		ret = ata_eh_get_ncq_success_sense(link);
2028
	else
2029
		ret = ata_eh_get_non_ncq_success_sense(link);
2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067
	if (ret)
		goto out;

	ata_eh_done(link, dev, ATA_EH_GET_SUCCESS_SENSE);
	return;

out:
	/*
	 * If we failed to get sense data for a successful command that ought to
	 * have sense data, we cannot simply return BLK_STS_OK to user space.
	 * This is because we can't know if the sense data that we couldn't get
	 * was actually "DATA CURRENTLY UNAVAILABLE". Reporting such a command
	 * as success to user space would result in a silent data corruption.
	 * Thus, add a bogus ABORTED_COMMAND sense data to such commands, such
	 * that SCSI will report these commands as BLK_STS_IOERR to user space.
	 */
	ata_qc_for_each_raw(ap, qc, tag) {
		if (!(qc->flags & ATA_QCFLAG_EH) ||
		    !(qc->flags & ATA_QCFLAG_EH_SUCCESS_CMD) ||
		    qc->err_mask ||
		    ata_dev_phys_link(qc->dev) != link)
			continue;

		/* We managed to get sense for this success command, skip. */
		if (qc->flags & ATA_QCFLAG_SENSE_VALID)
			continue;

		/* This success command did not have any sense data, skip. */
		if (!(qc->result_tf.status & ATA_SENSE))
			continue;

		/* This success command had sense data, but we failed to get. */
		ata_scsi_set_sense(dev, qc->scsicmd, ABORTED_COMMAND, 0, 0);
		qc->flags |= ATA_QCFLAG_SENSE_VALID;
	}
	ata_eh_done(link, dev, ATA_EH_GET_SUCCESS_SENSE);
}

2068
/**
2069 2070
 *	ata_eh_link_autopsy - analyze error and determine recovery action
 *	@link: host link to perform autopsy on
2071
 *
2072 2073 2074
 *	Analyze why @link failed and determine which recovery actions
 *	are needed.  This function also sets more detailed AC_ERR_*
 *	values and fills sense data for ATAPI CHECK SENSE.
2075 2076 2077 2078
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 */
2079
static void ata_eh_link_autopsy(struct ata_link *link)
2080
{
2081
	struct ata_port *ap = link->ap;
2082
	struct ata_eh_context *ehc = &link->eh_context;
2083
	struct ata_queued_cmd *qc;
2084
	struct ata_device *dev;
2085
	unsigned int all_err_mask = 0, eflags = 0;
2086
	int tag, nr_failed = 0, nr_quiet = 0;
2087 2088 2089
	u32 serror;
	int rc;

2090 2091 2092
	if (ehc->i.flags & ATA_EHI_NO_AUTOPSY)
		return;

2093
	/* obtain and analyze SError */
2094
	rc = sata_scr_read(link, SCR_ERROR, &serror);
2095 2096
	if (rc == 0) {
		ehc->i.serror |= serror;
2097
		ata_eh_analyze_serror(link);
2098
	} else if (rc != -EOPNOTSUPP) {
Tejun Heo's avatar
Tejun Heo committed
2099
		/* SError read failed, force reset and probing */
2100
		ehc->i.probe_mask |= ATA_ALL_DEVICES;
Tejun Heo's avatar
Tejun Heo committed
2101
		ehc->i.action |= ATA_EH_RESET;
2102 2103
		ehc->i.err_mask |= AC_ERR_OTHER;
	}
2104

2105
	/* analyze NCQ failure */
2106
	ata_eh_analyze_ncq_error(link);
2107

2108 2109 2110 2111 2112 2113 2114 2115
	/*
	 * Check if this was a successful command that simply needs sense data.
	 * Since the sense data is not part of the completion, we need to fetch
	 * it using an additional command. Since this can't be done from irq
	 * context, the sense data for successful commands are fetched by EH.
	 */
	ata_eh_get_success_sense(link);

2116 2117 2118 2119 2120 2121
	/* any real error trumps AC_ERR_OTHER */
	if (ehc->i.err_mask & ~AC_ERR_OTHER)
		ehc->i.err_mask &= ~AC_ERR_OTHER;

	all_err_mask |= ehc->i.err_mask;

2122
	ata_qc_for_each_raw(ap, qc, tag) {
2123
		if (!(qc->flags & ATA_QCFLAG_EH) ||
2124
		    qc->flags & ATA_QCFLAG_RETRY ||
2125
		    qc->flags & ATA_QCFLAG_EH_SUCCESS_CMD ||
Tejun Heo's avatar
Tejun Heo committed
2126
		    ata_dev_phys_link(qc->dev) != link)
2127 2128 2129 2130 2131 2132
			continue;

		/* inherit upper level err_mask */
		qc->err_mask |= ehc->i.err_mask;

		/* analyze TF */
2133
		ehc->i.action |= ata_eh_analyze_tf(qc);
2134 2135 2136 2137 2138 2139 2140 2141 2142 2143

		/* DEV errors are probably spurious in case of ATA_BUS error */
		if (qc->err_mask & AC_ERR_ATA_BUS)
			qc->err_mask &= ~(AC_ERR_DEV | AC_ERR_MEDIA |
					  AC_ERR_INVALID);

		/* any real error trumps unknown error */
		if (qc->err_mask & ~AC_ERR_OTHER)
			qc->err_mask &= ~AC_ERR_OTHER;

2144 2145 2146 2147 2148 2149 2150
		/*
		 * SENSE_VALID trumps dev/unknown error and revalidation. Upper
		 * layers will determine whether the command is worth retrying
		 * based on the sense data and device class/type. Otherwise,
		 * determine directly if the command is worth retrying using its
		 * error mask and flags.
		 */
2151
		if (qc->flags & ATA_QCFLAG_SENSE_VALID)
2152
			qc->err_mask &= ~(AC_ERR_DEV | AC_ERR_OTHER);
2153
		else if (ata_eh_worth_retry(qc))
2154 2155
			qc->flags |= ATA_QCFLAG_RETRY;

2156
		/* accumulate error info */
2157
		ehc->i.dev = qc->dev;
2158 2159
		all_err_mask |= qc->err_mask;
		if (qc->flags & ATA_QCFLAG_IO)
2160
			eflags |= ATA_EFLAG_IS_IO;
Hannes Reinecke's avatar
Hannes Reinecke committed
2161
		trace_ata_eh_link_autopsy_qc(qc);
2162 2163 2164 2165 2166

		/* Count quiet errors */
		if (ata_eh_quiet(qc))
			nr_quiet++;
		nr_failed++;
2167 2168
	}

2169 2170 2171 2172
	/* If all failed commands requested silence, then be quiet */
	if (nr_quiet == nr_failed)
		ehc->i.flags |= ATA_EHI_QUIET;

2173
	/* enforce default EH actions */
2174
	if (ata_port_is_frozen(ap) ||
2175
	    all_err_mask & (AC_ERR_HSM | AC_ERR_TIMEOUT))
Tejun Heo's avatar
Tejun Heo committed
2176
		ehc->i.action |= ATA_EH_RESET;
2177 2178
	else if (((eflags & ATA_EFLAG_IS_IO) && all_err_mask) ||
		 (!(eflags & ATA_EFLAG_IS_IO) && (all_err_mask & ~AC_ERR_DEV)))
2179
		ehc->i.action |= ATA_EH_REVALIDATE;
2180

2181 2182 2183
	/* If we have offending qcs and the associated failed device,
	 * perform per-dev EH action only on the offending device.
	 */
2184 2185 2186 2187
	if (ehc->i.dev) {
		ehc->i.dev_action[ehc->i.dev->devno] |=
			ehc->i.action & ATA_EH_PERDEV_MASK;
		ehc->i.action &= ~ATA_EH_PERDEV_MASK;
2188 2189
	}

2190 2191 2192 2193 2194
	/* propagate timeout to host link */
	if ((all_err_mask & AC_ERR_TIMEOUT) && !ata_is_host_link(link))
		ap->link.eh_context.i.err_mask |= AC_ERR_TIMEOUT;

	/* record error and consider speeding down */
2195
	dev = ehc->i.dev;
2196 2197 2198
	if (!dev && ((ata_link_max_devices(link) == 1 &&
		      ata_dev_enabled(link->device))))
	    dev = link->device;
2199

2200 2201 2202
	if (dev) {
		if (dev->flags & ATA_DFLAG_DUBIOUS_XFER)
			eflags |= ATA_EFLAG_DUBIOUS_XFER;
2203
		ehc->i.action |= ata_eh_speed_down(dev, eflags, all_err_mask);
2204
		trace_ata_eh_link_autopsy(dev, ehc->i.action, all_err_mask);
2205
	}
2206 2207 2208
}

/**
2209 2210 2211 2212 2213 2214 2215 2216 2217
 *	ata_eh_autopsy - analyze error and determine recovery action
 *	@ap: host port to perform autopsy on
 *
 *	Analyze all links of @ap and determine why they failed and
 *	which recovery actions are needed.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 */
2218
void ata_eh_autopsy(struct ata_port *ap)
2219 2220 2221
{
	struct ata_link *link;

Tejun Heo's avatar
Tejun Heo committed
2222
	ata_for_each_link(link, ap, EDGE)
2223
		ata_eh_link_autopsy(link);
2224

Tejun Heo's avatar
Tejun Heo committed
2225 2226 2227 2228 2229 2230 2231 2232
	/* Handle the frigging slave link.  Autopsy is done similarly
	 * but actions and flags are transferred over to the master
	 * link and handled from there.
	 */
	if (ap->slave_link) {
		struct ata_eh_context *mehc = &ap->link.eh_context;
		struct ata_eh_context *sehc = &ap->slave_link->eh_context;

2233 2234 2235 2236
		/* transfer control flags from master to slave */
		sehc->i.flags |= mehc->i.flags & ATA_EHI_TO_SLAVE_MASK;

		/* perform autopsy on the slave link */
Tejun Heo's avatar
Tejun Heo committed
2237 2238
		ata_eh_link_autopsy(ap->slave_link);

2239
		/* transfer actions from slave to master and clear slave */
Tejun Heo's avatar
Tejun Heo committed
2240 2241 2242 2243 2244 2245 2246
		ata_eh_about_to_do(ap->slave_link, NULL, ATA_EH_ALL_ACTIONS);
		mehc->i.action		|= sehc->i.action;
		mehc->i.dev_action[1]	|= sehc->i.dev_action[1];
		mehc->i.flags		|= sehc->i.flags;
		ata_eh_done(ap->slave_link, NULL, ATA_EH_ALL_ACTIONS);
	}

2247 2248 2249
	/* Autopsy of fanout ports can affect host link autopsy.
	 * Perform host link autopsy last.
	 */
Tejun Heo's avatar
Tejun Heo committed
2250
	if (sata_pmp_attached(ap))
2251
		ata_eh_link_autopsy(&ap->link);
2252 2253
}

2254
/**
2255 2256
 *	ata_get_cmd_name - get name for ATA command
 *	@command: ATA command code to get name for
2257
 *
2258
 *	Return a textual name of the given command or "unknown"
2259 2260 2261 2262
 *
 *	LOCKING:
 *	None
 */
2263
const char *ata_get_cmd_name(u8 command)
2264 2265 2266 2267 2268 2269 2270 2271
{
#ifdef CONFIG_ATA_VERBOSE_ERROR
	static const struct
	{
		u8 command;
		const char *text;
	} cmd_descr[] = {
		{ ATA_CMD_DEV_RESET,		"DEVICE RESET" },
2272 2273 2274 2275 2276
		{ ATA_CMD_CHK_POWER,		"CHECK POWER MODE" },
		{ ATA_CMD_STANDBY,		"STANDBY" },
		{ ATA_CMD_IDLE,			"IDLE" },
		{ ATA_CMD_EDD,			"EXECUTE DEVICE DIAGNOSTIC" },
		{ ATA_CMD_DOWNLOAD_MICRO,	"DOWNLOAD MICROCODE" },
2277
		{ ATA_CMD_DOWNLOAD_MICRO_DMA,	"DOWNLOAD MICROCODE DMA" },
2278
		{ ATA_CMD_NOP,			"NOP" },
2279 2280 2281 2282 2283 2284 2285 2286 2287
		{ ATA_CMD_FLUSH,		"FLUSH CACHE" },
		{ ATA_CMD_FLUSH_EXT,		"FLUSH CACHE EXT" },
		{ ATA_CMD_ID_ATA,		"IDENTIFY DEVICE" },
		{ ATA_CMD_ID_ATAPI,		"IDENTIFY PACKET DEVICE" },
		{ ATA_CMD_SERVICE,		"SERVICE" },
		{ ATA_CMD_READ,			"READ DMA" },
		{ ATA_CMD_READ_EXT,		"READ DMA EXT" },
		{ ATA_CMD_READ_QUEUED,		"READ DMA QUEUED" },
		{ ATA_CMD_READ_STREAM_EXT,	"READ STREAM EXT" },
2288
		{ ATA_CMD_READ_STREAM_DMA_EXT,  "READ STREAM DMA EXT" },
2289 2290 2291 2292
		{ ATA_CMD_WRITE,		"WRITE DMA" },
		{ ATA_CMD_WRITE_EXT,		"WRITE DMA EXT" },
		{ ATA_CMD_WRITE_QUEUED,		"WRITE DMA QUEUED EXT" },
		{ ATA_CMD_WRITE_STREAM_EXT,	"WRITE STREAM EXT" },
2293 2294 2295 2296 2297
		{ ATA_CMD_WRITE_STREAM_DMA_EXT, "WRITE STREAM DMA EXT" },
		{ ATA_CMD_WRITE_FUA_EXT,	"WRITE DMA FUA EXT" },
		{ ATA_CMD_WRITE_QUEUED_FUA_EXT, "WRITE DMA QUEUED FUA EXT" },
		{ ATA_CMD_FPDMA_READ,		"READ FPDMA QUEUED" },
		{ ATA_CMD_FPDMA_WRITE,		"WRITE FPDMA QUEUED" },
2298
		{ ATA_CMD_NCQ_NON_DATA,		"NCQ NON-DATA" },
2299 2300
		{ ATA_CMD_FPDMA_SEND,		"SEND FPDMA QUEUED" },
		{ ATA_CMD_FPDMA_RECV,		"RECEIVE FPDMA QUEUED" },
2301 2302 2303 2304 2305 2306 2307 2308
		{ ATA_CMD_PIO_READ,		"READ SECTOR(S)" },
		{ ATA_CMD_PIO_READ_EXT,		"READ SECTOR(S) EXT" },
		{ ATA_CMD_PIO_WRITE,		"WRITE SECTOR(S)" },
		{ ATA_CMD_PIO_WRITE_EXT,	"WRITE SECTOR(S) EXT" },
		{ ATA_CMD_READ_MULTI,		"READ MULTIPLE" },
		{ ATA_CMD_READ_MULTI_EXT,	"READ MULTIPLE EXT" },
		{ ATA_CMD_WRITE_MULTI,		"WRITE MULTIPLE" },
		{ ATA_CMD_WRITE_MULTI_EXT,	"WRITE MULTIPLE EXT" },
2309
		{ ATA_CMD_WRITE_MULTI_FUA_EXT,	"WRITE MULTIPLE FUA EXT" },
2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325
		{ ATA_CMD_SET_FEATURES,		"SET FEATURES" },
		{ ATA_CMD_SET_MULTI,		"SET MULTIPLE MODE" },
		{ ATA_CMD_VERIFY,		"READ VERIFY SECTOR(S)" },
		{ ATA_CMD_VERIFY_EXT,		"READ VERIFY SECTOR(S) EXT" },
		{ ATA_CMD_WRITE_UNCORR_EXT,	"WRITE UNCORRECTABLE EXT" },
		{ ATA_CMD_STANDBYNOW1,		"STANDBY IMMEDIATE" },
		{ ATA_CMD_IDLEIMMEDIATE,	"IDLE IMMEDIATE" },
		{ ATA_CMD_SLEEP,		"SLEEP" },
		{ ATA_CMD_INIT_DEV_PARAMS,	"INITIALIZE DEVICE PARAMETERS" },
		{ ATA_CMD_READ_NATIVE_MAX,	"READ NATIVE MAX ADDRESS" },
		{ ATA_CMD_READ_NATIVE_MAX_EXT,	"READ NATIVE MAX ADDRESS EXT" },
		{ ATA_CMD_SET_MAX,		"SET MAX ADDRESS" },
		{ ATA_CMD_SET_MAX_EXT,		"SET MAX ADDRESS EXT" },
		{ ATA_CMD_READ_LOG_EXT,		"READ LOG EXT" },
		{ ATA_CMD_WRITE_LOG_EXT,	"WRITE LOG EXT" },
		{ ATA_CMD_READ_LOG_DMA_EXT,	"READ LOG DMA EXT" },
2326
		{ ATA_CMD_WRITE_LOG_DMA_EXT,	"WRITE LOG DMA EXT" },
2327
		{ ATA_CMD_TRUSTED_NONDATA,	"TRUSTED NON-DATA" },
2328
		{ ATA_CMD_TRUSTED_RCV,		"TRUSTED RECEIVE" },
2329
		{ ATA_CMD_TRUSTED_RCV_DMA,	"TRUSTED RECEIVE DMA" },
2330
		{ ATA_CMD_TRUSTED_SND,		"TRUSTED SEND" },
2331
		{ ATA_CMD_TRUSTED_SND_DMA,	"TRUSTED SEND DMA" },
2332
		{ ATA_CMD_PMP_READ,		"READ BUFFER" },
2333
		{ ATA_CMD_PMP_READ_DMA,		"READ BUFFER DMA" },
2334
		{ ATA_CMD_PMP_WRITE,		"WRITE BUFFER" },
2335
		{ ATA_CMD_PMP_WRITE_DMA,	"WRITE BUFFER DMA" },
2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346
		{ ATA_CMD_CONF_OVERLAY,		"DEVICE CONFIGURATION OVERLAY" },
		{ ATA_CMD_SEC_SET_PASS,		"SECURITY SET PASSWORD" },
		{ ATA_CMD_SEC_UNLOCK,		"SECURITY UNLOCK" },
		{ ATA_CMD_SEC_ERASE_PREP,	"SECURITY ERASE PREPARE" },
		{ ATA_CMD_SEC_ERASE_UNIT,	"SECURITY ERASE UNIT" },
		{ ATA_CMD_SEC_FREEZE_LOCK,	"SECURITY FREEZE LOCK" },
		{ ATA_CMD_SEC_DISABLE_PASS,	"SECURITY DISABLE PASSWORD" },
		{ ATA_CMD_CONFIG_STREAM,	"CONFIGURE STREAM" },
		{ ATA_CMD_SMART,		"SMART" },
		{ ATA_CMD_MEDIA_LOCK,		"DOOR LOCK" },
		{ ATA_CMD_MEDIA_UNLOCK,		"DOOR UNLOCK" },
2347
		{ ATA_CMD_DSM,			"DATA SET MANAGEMENT" },
2348 2349
		{ ATA_CMD_CHK_MED_CRD_TYP,	"CHECK MEDIA CARD TYPE" },
		{ ATA_CMD_CFA_REQ_EXT_ERR,	"CFA REQUEST EXTENDED ERROR" },
2350 2351 2352
		{ ATA_CMD_CFA_WRITE_NE,		"CFA WRITE SECTORS WITHOUT ERASE" },
		{ ATA_CMD_CFA_TRANS_SECT,	"CFA TRANSLATE SECTOR" },
		{ ATA_CMD_CFA_ERASE,		"CFA ERASE SECTORS" },
2353
		{ ATA_CMD_CFA_WRITE_MULT_NE,	"CFA WRITE MULTIPLE WITHOUT ERASE" },
2354 2355
		{ ATA_CMD_REQ_SENSE_DATA,	"REQUEST SENSE DATA EXT" },
		{ ATA_CMD_SANITIZE_DEVICE,	"SANITIZE DEVICE" },
2356
		{ ATA_CMD_ZAC_MGMT_IN,		"ZAC MANAGEMENT IN" },
2357
		{ ATA_CMD_ZAC_MGMT_OUT,		"ZAC MANAGEMENT OUT" },
2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371
		{ ATA_CMD_READ_LONG,		"READ LONG (with retries)" },
		{ ATA_CMD_READ_LONG_ONCE,	"READ LONG (without retries)" },
		{ ATA_CMD_WRITE_LONG,		"WRITE LONG (with retries)" },
		{ ATA_CMD_WRITE_LONG_ONCE,	"WRITE LONG (without retries)" },
		{ ATA_CMD_RESTORE,		"RECALIBRATE" },
		{ 0,				NULL } /* terminate list */
	};

	unsigned int i;
	for (i = 0; cmd_descr[i].text; i++)
		if (cmd_descr[i].command == command)
			return cmd_descr[i].text;
#endif

2372
	return "unknown";
2373
}
2374
EXPORT_SYMBOL_GPL(ata_get_cmd_name);
2375

2376 2377
/**
 *	ata_eh_link_report - report error handling to user
2378
 *	@link: ATA link EH is going on
2379 2380 2381 2382 2383 2384
 *
 *	Report EH to user.
 *
 *	LOCKING:
 *	None.
 */
2385
static void ata_eh_link_report(struct ata_link *link)
2386
{
2387 2388
	struct ata_port *ap = link->ap;
	struct ata_eh_context *ehc = &link->eh_context;
2389
	struct ata_queued_cmd *qc;
2390
	const char *frozen, *desc;
2391
	char tries_buf[16] = "";
2392 2393
	int tag, nr_failed = 0;

2394 2395 2396
	if (ehc->i.flags & ATA_EHI_QUIET)
		return;

2397 2398 2399 2400
	desc = NULL;
	if (ehc->i.desc[0] != '\0')
		desc = ehc->i.desc;

2401
	ata_qc_for_each_raw(ap, qc, tag) {
2402
		if (!(qc->flags & ATA_QCFLAG_EH) ||
Tejun Heo's avatar
Tejun Heo committed
2403
		    ata_dev_phys_link(qc->dev) != link ||
2404 2405
		    ((qc->flags & ATA_QCFLAG_QUIET) &&
		     qc->err_mask == AC_ERR_DEV))
2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416
			continue;
		if (qc->flags & ATA_QCFLAG_SENSE_VALID && !qc->err_mask)
			continue;

		nr_failed++;
	}

	if (!nr_failed && !ehc->i.err_mask)
		return;

	frozen = "";
2417
	if (ata_port_is_frozen(ap))
2418 2419
		frozen = " frozen";

2420
	if (ap->eh_tries < ATA_EH_MAX_TRIES)
2421
		snprintf(tries_buf, sizeof(tries_buf), " t%d",
2422 2423
			 ap->eh_tries);

2424
	if (ehc->i.dev) {
2425 2426 2427 2428
		ata_dev_err(ehc->i.dev, "exception Emask 0x%x "
			    "SAct 0x%x SErr 0x%x action 0x%x%s%s\n",
			    ehc->i.err_mask, link->sactive, ehc->i.serror,
			    ehc->i.action, frozen, tries_buf);
2429
		if (desc)
2430
			ata_dev_err(ehc->i.dev, "%s\n", desc);
2431
	} else {
2432 2433 2434 2435
		ata_link_err(link, "exception Emask 0x%x "
			     "SAct 0x%x SErr 0x%x action 0x%x%s%s\n",
			     ehc->i.err_mask, link->sactive, ehc->i.serror,
			     ehc->i.action, frozen, tries_buf);
2436
		if (desc)
2437
			ata_link_err(link, "%s\n", desc);
2438 2439
	}

2440
#ifdef CONFIG_ATA_VERBOSE_ERROR
2441
	if (ehc->i.serror)
2442
		ata_link_err(link,
2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459
		  "SError: { %s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s}\n",
		  ehc->i.serror & SERR_DATA_RECOVERED ? "RecovData " : "",
		  ehc->i.serror & SERR_COMM_RECOVERED ? "RecovComm " : "",
		  ehc->i.serror & SERR_DATA ? "UnrecovData " : "",
		  ehc->i.serror & SERR_PERSISTENT ? "Persist " : "",
		  ehc->i.serror & SERR_PROTOCOL ? "Proto " : "",
		  ehc->i.serror & SERR_INTERNAL ? "HostInt " : "",
		  ehc->i.serror & SERR_PHYRDY_CHG ? "PHYRdyChg " : "",
		  ehc->i.serror & SERR_PHY_INT_ERR ? "PHYInt " : "",
		  ehc->i.serror & SERR_COMM_WAKE ? "CommWake " : "",
		  ehc->i.serror & SERR_10B_8B_ERR ? "10B8B " : "",
		  ehc->i.serror & SERR_DISPARITY ? "Dispar " : "",
		  ehc->i.serror & SERR_CRC ? "BadCRC " : "",
		  ehc->i.serror & SERR_HANDSHAKE ? "Handshk " : "",
		  ehc->i.serror & SERR_LINK_SEQ_ERR ? "LinkSeq " : "",
		  ehc->i.serror & SERR_TRANS_ST_ERROR ? "TrStaTrns " : "",
		  ehc->i.serror & SERR_UNRECOG_FIS ? "UnrecFIS " : "",
2460
		  ehc->i.serror & SERR_DEV_XCHG ? "DevExch " : "");
2461
#endif
2462

2463
	ata_qc_for_each_raw(ap, qc, tag) {
2464
		struct ata_taskfile *cmd = &qc->tf, *res = &qc->result_tf;
2465 2466
		char data_buf[20] = "";
		char cdb_buf[70] = "";
2467

2468
		if (!(qc->flags & ATA_QCFLAG_EH) ||
Tejun Heo's avatar
Tejun Heo committed
2469
		    ata_dev_phys_link(qc->dev) != link || !qc->err_mask)
2470 2471
			continue;

2472 2473 2474 2475 2476 2477
		if (qc->dma_dir != DMA_NONE) {
			static const char *dma_str[] = {
				[DMA_BIDIRECTIONAL]	= "bidi",
				[DMA_TO_DEVICE]		= "out",
				[DMA_FROM_DEVICE]	= "in",
			};
2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508
			const char *prot_str = NULL;

			switch (qc->tf.protocol) {
			case ATA_PROT_UNKNOWN:
				prot_str = "unknown";
				break;
			case ATA_PROT_NODATA:
				prot_str = "nodata";
				break;
			case ATA_PROT_PIO:
				prot_str = "pio";
				break;
			case ATA_PROT_DMA:
				prot_str = "dma";
				break;
			case ATA_PROT_NCQ:
				prot_str = "ncq dma";
				break;
			case ATA_PROT_NCQ_NODATA:
				prot_str = "ncq nodata";
				break;
			case ATAPI_PROT_NODATA:
				prot_str = "nodata";
				break;
			case ATAPI_PROT_PIO:
				prot_str = "pio";
				break;
			case ATAPI_PROT_DMA:
				prot_str = "dma";
				break;
			}
2509
			snprintf(data_buf, sizeof(data_buf), " %s %u %s",
2510
				 prot_str, qc->nbytes, dma_str[qc->dma_dir]);
2511 2512
		}

2513
		if (ata_is_atapi(qc->tf.protocol)) {
2514 2515 2516
			const u8 *cdb = qc->cdb;
			size_t cdb_len = qc->dev->cdb_len;

2517 2518 2519 2520 2521 2522
			if (qc->scsicmd) {
				cdb = qc->scsicmd->cmnd;
				cdb_len = qc->scsicmd->cmd_len;
			}
			__scsi_format_command(cdb_buf, sizeof(cdb_buf),
					      cdb, cdb_len);
2523 2524 2525
		} else
			ata_dev_err(qc->dev, "failed command: %s\n",
				    ata_get_cmd_name(cmd->command));
2526

2527
		ata_dev_err(qc->dev,
2528
			"cmd %02x/%02x:%02x:%02x:%02x:%02x/%02x:%02x:%02x:%02x:%02x/%02x "
2529
			"tag %d%s\n         %s"
2530
			"res %02x/%02x:%02x:%02x:%02x:%02x/%02x:%02x:%02x:%02x:%02x/%02x "
Tejun Heo's avatar
Tejun Heo committed
2531
			"Emask 0x%x (%s)%s\n",
2532 2533 2534 2535
			cmd->command, cmd->feature, cmd->nsect,
			cmd->lbal, cmd->lbam, cmd->lbah,
			cmd->hob_feature, cmd->hob_nsect,
			cmd->hob_lbal, cmd->hob_lbam, cmd->hob_lbah,
2536
			cmd->device, qc->tag, data_buf, cdb_buf,
2537
			res->status, res->error, res->nsect,
2538 2539 2540
			res->lbal, res->lbam, res->lbah,
			res->hob_feature, res->hob_nsect,
			res->hob_lbal, res->hob_lbam, res->hob_lbah,
Tejun Heo's avatar
Tejun Heo committed
2541 2542
			res->device, qc->err_mask, ata_err_string(qc->err_mask),
			qc->err_mask & AC_ERR_NCQ ? " <F>" : "");
2543

2544
#ifdef CONFIG_ATA_VERBOSE_ERROR
2545 2546 2547
		if (res->status & (ATA_BUSY | ATA_DRDY | ATA_DF | ATA_DRQ |
				   ATA_SENSE | ATA_ERR)) {
			if (res->status & ATA_BUSY)
2548
				ata_dev_err(qc->dev, "status: { Busy }\n");
2549
			else
2550
				ata_dev_err(qc->dev, "status: { %s%s%s%s%s}\n",
2551 2552 2553 2554 2555
				  res->status & ATA_DRDY ? "DRDY " : "",
				  res->status & ATA_DF ? "DF " : "",
				  res->status & ATA_DRQ ? "DRQ " : "",
				  res->status & ATA_SENSE ? "SENSE " : "",
				  res->status & ATA_ERR ? "ERR " : "");
2556 2557 2558
		}

		if (cmd->command != ATA_CMD_PACKET &&
2559 2560
		    (res->error & (ATA_ICRC | ATA_UNC | ATA_AMNF | ATA_IDNF |
				   ATA_ABORTED)))
2561
			ata_dev_err(qc->dev, "error: { %s%s%s%s%s}\n",
2562 2563 2564 2565 2566
				    res->error & ATA_ICRC ? "ICRC " : "",
				    res->error & ATA_UNC ? "UNC " : "",
				    res->error & ATA_AMNF ? "AMNF " : "",
				    res->error & ATA_IDNF ? "IDNF " : "",
				    res->error & ATA_ABORTED ? "ABRT " : "");
2567
#endif
2568 2569 2570
	}
}

2571 2572 2573 2574 2575 2576 2577 2578 2579
/**
 *	ata_eh_report - report error handling to user
 *	@ap: ATA port to report EH about
 *
 *	Report EH to user.
 *
 *	LOCKING:
 *	None.
 */
2580
void ata_eh_report(struct ata_port *ap)
2581 2582 2583
{
	struct ata_link *link;

Tejun Heo's avatar
Tejun Heo committed
2584
	ata_for_each_link(link, ap, HOST_FIRST)
2585 2586 2587
		ata_eh_link_report(link);
}

Tejun Heo's avatar
Tejun Heo committed
2588
static int ata_do_reset(struct ata_link *link, ata_reset_fn_t reset,
Tejun Heo's avatar
Tejun Heo committed
2589 2590
			unsigned int *classes, unsigned long deadline,
			bool clear_classes)
2591
{
2592
	struct ata_device *dev;
2593

Tejun Heo's avatar
Tejun Heo committed
2594
	if (clear_classes)
Tejun Heo's avatar
Tejun Heo committed
2595
		ata_for_each_dev(dev, link, ALL)
Tejun Heo's avatar
Tejun Heo committed
2596
			classes[dev->devno] = ATA_DEV_UNKNOWN;
2597

2598
	return reset(link, classes, deadline);
2599 2600
}

2601
static int ata_eh_followup_srst_needed(struct ata_link *link, int rc)
2602
{
2603
	if ((link->flags & ATA_LFLAG_NO_SRST) || ata_link_offline(link))
2604
		return 0;
2605 2606
	if (rc == -EAGAIN)
		return 1;
Tejun Heo's avatar
Tejun Heo committed
2607
	if (sata_pmp_supported(link->ap) && ata_is_host_link(link))
2608
		return 1;
2609 2610 2611
	return 0;
}

2612 2613 2614
int ata_eh_reset(struct ata_link *link, int classify,
		 ata_prereset_fn_t prereset, ata_reset_fn_t softreset,
		 ata_reset_fn_t hardreset, ata_postreset_fn_t postreset)
2615
{
2616
	struct ata_port *ap = link->ap;
Tejun Heo's avatar
Tejun Heo committed
2617
	struct ata_link *slave = ap->slave_link;
2618
	struct ata_eh_context *ehc = &link->eh_context;
2619
	struct ata_eh_context *sehc = slave ? &slave->eh_context : NULL;
2620
	unsigned int *classes = ehc->classes;
2621
	unsigned int lflags = link->flags;
2622
	int verbose = !(ehc->i.flags & ATA_EHI_QUIET);
2623
	int max_tries = 0, try = 0;
Tejun Heo's avatar
Tejun Heo committed
2624
	struct ata_link *failed_link;
2625
	struct ata_device *dev;
2626
	unsigned long deadline, now;
2627
	ata_reset_fn_t reset;
2628
	unsigned long flags;
2629
	u32 sstatus;
Tejun Heo's avatar
Tejun Heo committed
2630
	int nr_unknown, rc;
2631

2632 2633 2634
	/*
	 * Prepare to reset
	 */
2635
	while (ata_eh_reset_timeouts[max_tries] != UINT_MAX)
2636
		max_tries++;
Dan Williams's avatar
Dan Williams committed
2637 2638
	if (link->flags & ATA_LFLAG_RST_ONCE)
		max_tries = 1;
2639 2640 2641 2642
	if (link->flags & ATA_LFLAG_NO_HRST)
		hardreset = NULL;
	if (link->flags & ATA_LFLAG_NO_SRST)
		softreset = NULL;
2643

Lucas De Marchi's avatar
Lucas De Marchi committed
2644
	/* make sure each reset attempt is at least COOL_DOWN apart */
2645 2646 2647 2648 2649 2650 2651 2652
	if (ehc->i.flags & ATA_EHI_DID_RESET) {
		now = jiffies;
		WARN_ON(time_after(ehc->last_reset, now));
		deadline = ata_deadline(ehc->last_reset,
					ATA_EH_RESET_COOL_DOWN);
		if (time_before(now, deadline))
			schedule_timeout_uninterruptible(deadline - now);
	}
2653

2654 2655 2656 2657
	spin_lock_irqsave(ap->lock, flags);
	ap->pflags |= ATA_PFLAG_RESETTING;
	spin_unlock_irqrestore(ap->lock, flags);

Tejun Heo's avatar
Tejun Heo committed
2658
	ata_eh_about_to_do(link, NULL, ATA_EH_RESET);
2659

Tejun Heo's avatar
Tejun Heo committed
2660
	ata_for_each_dev(dev, link, ALL) {
2661 2662 2663 2664 2665 2666 2667 2668
		/* If we issue an SRST then an ATA drive (not ATAPI)
		 * may change configuration and be in PIO0 timing. If
		 * we do a hard reset (or are coming from power on)
		 * this is true for ATA or ATAPI. Until we've set a
		 * suitable controller mode we should not touch the
		 * bus as we may be talking too fast.
		 */
		dev->pio_mode = XFER_PIO_0;
2669
		dev->dma_mode = 0xff;
2670 2671 2672 2673 2674 2675 2676 2677 2678 2679

		/* If the controller has a pio mode setup function
		 * then use it to set the chipset to rights. Don't
		 * touch the DMA setup as that will be dealt with when
		 * configuring devices.
		 */
		if (ap->ops->set_piomode)
			ap->ops->set_piomode(ap, dev);
	}

Tejun Heo's avatar
Tejun Heo committed
2680
	/* prefer hardreset */
2681
	reset = NULL;
Tejun Heo's avatar
Tejun Heo committed
2682 2683 2684
	ehc->i.action &= ~ATA_EH_RESET;
	if (hardreset) {
		reset = hardreset;
2685
		ehc->i.action |= ATA_EH_HARDRESET;
2686
	} else if (softreset) {
Tejun Heo's avatar
Tejun Heo committed
2687
		reset = softreset;
2688
		ehc->i.action |= ATA_EH_SOFTRESET;
Tejun Heo's avatar
Tejun Heo committed
2689
	}
2690 2691

	if (prereset) {
Tejun Heo's avatar
Tejun Heo committed
2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715
		unsigned long deadline = ata_deadline(jiffies,
						      ATA_EH_PRERESET_TIMEOUT);

		if (slave) {
			sehc->i.action &= ~ATA_EH_RESET;
			sehc->i.action |= ehc->i.action;
		}

		rc = prereset(link, deadline);

		/* If present, do prereset on slave link too.  Reset
		 * is skipped iff both master and slave links report
		 * -ENOENT or clear ATA_EH_RESET.
		 */
		if (slave && (rc == 0 || rc == -ENOENT)) {
			int tmp;

			tmp = prereset(slave, deadline);
			if (tmp != -ENOENT)
				rc = tmp;

			ehc->i.action |= sehc->i.action;
		}

2716
		if (rc) {
2717
			if (rc == -ENOENT) {
2718
				ata_link_dbg(link, "port disabled--ignoring\n");
Tejun Heo's avatar
Tejun Heo committed
2719
				ehc->i.action &= ~ATA_EH_RESET;
2720

Tejun Heo's avatar
Tejun Heo committed
2721
				ata_for_each_dev(dev, link, ALL)
2722
					classes[dev->devno] = ATA_DEV_NONE;
2723 2724

				rc = 0;
2725
			} else
2726 2727 2728
				ata_link_err(link,
					     "prereset failed (errno=%d)\n",
					     rc);
2729
			goto out;
2730 2731
		}

2732
		/* prereset() might have cleared ATA_EH_RESET.  If so,
2733
		 * bang classes, thaw and return.
2734 2735
		 */
		if (reset && !(ehc->i.action & ATA_EH_RESET)) {
Tejun Heo's avatar
Tejun Heo committed
2736
			ata_for_each_dev(dev, link, ALL)
2737
				classes[dev->devno] = ATA_DEV_NONE;
2738
			if (ata_port_is_frozen(ap) && ata_is_host_link(link))
2739
				ata_eh_thaw_port(ap);
2740 2741 2742
			rc = 0;
			goto out;
		}
2743 2744
	}

2745
 retry:
2746 2747 2748
	/*
	 * Perform reset
	 */
2749 2750 2751
	if (ata_is_host_link(link))
		ata_eh_freeze_port(ap);

2752
	deadline = ata_deadline(jiffies, ata_eh_reset_timeouts[try++]);
2753

2754 2755
	if (reset) {
		if (verbose)
2756 2757
			ata_link_info(link, "%s resetting link\n",
				      reset == softreset ? "soft" : "hard");
2758 2759

		/* mark that this EH session started with reset */
2760
		ehc->last_reset = jiffies;
2761
		if (reset == hardreset) {
2762
			ehc->i.flags |= ATA_EHI_DID_HARDRESET;
2763 2764
			trace_ata_link_hardreset_begin(link, classes, deadline);
		} else {
2765
			ehc->i.flags |= ATA_EHI_DID_SOFTRESET;
2766 2767
			trace_ata_link_softreset_begin(link, classes, deadline);
		}
2768

Tejun Heo's avatar
Tejun Heo committed
2769
		rc = ata_do_reset(link, reset, classes, deadline, true);
2770 2771 2772 2773
		if (reset == hardreset)
			trace_ata_link_hardreset_end(link, classes, rc);
		else
			trace_ata_link_softreset_end(link, classes, rc);
Tejun Heo's avatar
Tejun Heo committed
2774 2775
		if (rc && rc != -EAGAIN) {
			failed_link = link;
2776
			goto fail;
Tejun Heo's avatar
Tejun Heo committed
2777 2778 2779 2780 2781 2782 2783
		}

		/* hardreset slave link if existent */
		if (slave && reset == hardreset) {
			int tmp;

			if (verbose)
2784
				ata_link_info(slave, "hard resetting link\n");
Tejun Heo's avatar
Tejun Heo committed
2785 2786

			ata_eh_about_to_do(slave, NULL, ATA_EH_RESET);
2787 2788
			trace_ata_slave_hardreset_begin(slave, classes,
							deadline);
Tejun Heo's avatar
Tejun Heo committed
2789 2790
			tmp = ata_do_reset(slave, reset, classes, deadline,
					   false);
2791
			trace_ata_slave_hardreset_end(slave, classes, tmp);
Tejun Heo's avatar
Tejun Heo committed
2792 2793 2794
			switch (tmp) {
			case -EAGAIN:
				rc = -EAGAIN;
2795
				break;
Tejun Heo's avatar
Tejun Heo committed
2796 2797 2798 2799 2800 2801 2802 2803
			case 0:
				break;
			default:
				failed_link = slave;
				rc = tmp;
				goto fail;
			}
		}
2804

Tejun Heo's avatar
Tejun Heo committed
2805
		/* perform follow-up SRST if necessary */
2806
		if (reset == hardreset &&
2807
		    ata_eh_followup_srst_needed(link, rc)) {
2808
			reset = softreset;
2809

2810
			if (!reset) {
2811 2812
				ata_link_err(link,
	     "follow-up softreset required but no softreset available\n");
Tejun Heo's avatar
Tejun Heo committed
2813
				failed_link = link;
2814 2815 2816
				rc = -EINVAL;
				goto fail;
			}
2817

2818
			ata_eh_about_to_do(link, NULL, ATA_EH_RESET);
2819
			trace_ata_link_softreset_begin(link, classes, deadline);
Tejun Heo's avatar
Tejun Heo committed
2820
			rc = ata_do_reset(link, reset, classes, deadline, true);
2821
			trace_ata_link_softreset_end(link, classes, rc);
2822 2823 2824 2825
			if (rc) {
				failed_link = link;
				goto fail;
			}
2826
		}
2827 2828
	} else {
		if (verbose)
2829 2830
			ata_link_info(link,
	"no reset method available, skipping reset\n");
2831 2832
		if (!(lflags & ATA_LFLAG_ASSUME_CLASS))
			lflags |= ATA_LFLAG_ASSUME_ATA;
2833 2834
	}

2835 2836 2837
	/*
	 * Post-reset processing
	 */
Tejun Heo's avatar
Tejun Heo committed
2838
	ata_for_each_dev(dev, link, ALL) {
2839 2840 2841 2842 2843 2844
		/* After the reset, the device state is PIO 0 and the
		 * controller state is undefined.  Reset also wakes up
		 * drives from sleeping mode.
		 */
		dev->pio_mode = XFER_PIO_0;
		dev->flags &= ~ATA_DFLAG_SLEEPING;
2845

2846 2847 2848 2849 2850 2851 2852 2853
		if (ata_phys_link_offline(ata_dev_phys_link(dev)))
			continue;

		/* apply class override */
		if (lflags & ATA_LFLAG_ASSUME_ATA)
			classes[dev->devno] = ATA_DEV_ATA;
		else if (lflags & ATA_LFLAG_ASSUME_SEMB)
			classes[dev->devno] = ATA_DEV_SEMB_UNSUP;
2854 2855
	}

2856 2857 2858
	/* record current link speed */
	if (sata_scr_read(link, SCR_STATUS, &sstatus) == 0)
		link->sata_spd = (sstatus >> 4) & 0xf;
Tejun Heo's avatar
Tejun Heo committed
2859 2860
	if (slave && sata_scr_read(slave, SCR_STATUS, &sstatus) == 0)
		slave->sata_spd = (sstatus >> 4) & 0xf;
2861

2862 2863 2864 2865
	/* thaw the port */
	if (ata_is_host_link(link))
		ata_eh_thaw_port(ap);

2866 2867 2868 2869 2870 2871 2872
	/* postreset() should clear hardware SError.  Although SError
	 * is cleared during link resume, clearing SError here is
	 * necessary as some PHYs raise hotplug events after SRST.
	 * This introduces race condition where hotplug occurs between
	 * reset and here.  This race is mediated by cross checking
	 * link onlineness and classification result later.
	 */
Tejun Heo's avatar
Tejun Heo committed
2873
	if (postreset) {
2874
		postreset(link, classes);
2875 2876
		trace_ata_link_postreset(link, classes, rc);
		if (slave) {
Tejun Heo's avatar
Tejun Heo committed
2877
			postreset(slave, classes);
2878 2879
			trace_ata_slave_postreset(slave, classes, rc);
		}
Tejun Heo's avatar
Tejun Heo committed
2880
	}
2881

2882
	/* clear cached SError */
2883
	spin_lock_irqsave(link->ap->lock, flags);
2884
	link->eh_info.serror = 0;
Tejun Heo's avatar
Tejun Heo committed
2885
	if (slave)
2886
		slave->eh_info.serror = 0;
2887 2888
	spin_unlock_irqrestore(link->ap->lock, flags);

2889 2890
	/*
	 * Make sure onlineness and classification result correspond.
2891 2892 2893
	 * Hotplug could have happened during reset and some
	 * controllers fail to wait while a drive is spinning up after
	 * being hotplugged causing misdetection.  By cross checking
2894 2895
	 * link on/offlineness and classification result, those
	 * conditions can be reliably detected and retried.
2896
	 */
Tejun Heo's avatar
Tejun Heo committed
2897
	nr_unknown = 0;
Tejun Heo's avatar
Tejun Heo committed
2898
	ata_for_each_dev(dev, link, ALL) {
2899 2900
		if (ata_phys_link_online(ata_dev_phys_link(dev))) {
			if (classes[dev->devno] == ATA_DEV_UNKNOWN) {
2901
				ata_dev_dbg(dev, "link online but device misclassified\n");
2902
				classes[dev->devno] = ATA_DEV_NONE;
Tejun Heo's avatar
Tejun Heo committed
2903
				nr_unknown++;
2904 2905 2906
			}
		} else if (ata_phys_link_offline(ata_dev_phys_link(dev))) {
			if (ata_class_enabled(classes[dev->devno]))
2907 2908 2909
				ata_dev_dbg(dev,
					    "link offline, clearing class %d to NONE\n",
					    classes[dev->devno]);
2910 2911
			classes[dev->devno] = ATA_DEV_NONE;
		} else if (classes[dev->devno] == ATA_DEV_UNKNOWN) {
2912 2913
			ata_dev_dbg(dev,
				    "link status unknown, clearing UNKNOWN to NONE\n");
2914
			classes[dev->devno] = ATA_DEV_NONE;
Tejun Heo's avatar
Tejun Heo committed
2915
		}
2916 2917
	}

Tejun Heo's avatar
Tejun Heo committed
2918
	if (classify && nr_unknown) {
2919
		if (try < max_tries) {
2920 2921 2922
			ata_link_warn(link,
				      "link online but %d devices misclassified, retrying\n",
				      nr_unknown);
Tejun Heo's avatar
Tejun Heo committed
2923
			failed_link = link;
2924 2925 2926
			rc = -EAGAIN;
			goto fail;
		}
2927 2928 2929
		ata_link_warn(link,
			      "link online but %d devices misclassified, "
			      "device detection might fail\n", nr_unknown);
2930 2931
	}

2932
	/* reset successful, schedule revalidation */
Tejun Heo's avatar
Tejun Heo committed
2933
	ata_eh_done(link, NULL, ATA_EH_RESET);
Tejun Heo's avatar
Tejun Heo committed
2934 2935
	if (slave)
		ata_eh_done(slave, NULL, ATA_EH_RESET);
2936
	ehc->last_reset = jiffies;		/* update to completion time */
2937
	ehc->i.action |= ATA_EH_REVALIDATE;
2938
	link->lpm_policy = ATA_LPM_UNKNOWN;	/* reset LPM state */
2939

2940
	rc = 0;
2941 2942 2943
 out:
	/* clear hotplug flag */
	ehc->i.flags &= ~ATA_EHI_HOTPLUGGED;
Tejun Heo's avatar
Tejun Heo committed
2944 2945
	if (slave)
		sehc->i.flags &= ~ATA_EHI_HOTPLUGGED;
2946 2947 2948 2949 2950

	spin_lock_irqsave(ap->lock, flags);
	ap->pflags &= ~ATA_PFLAG_RESETTING;
	spin_unlock_irqrestore(ap->lock, flags);

2951
	return rc;
2952 2953

 fail:
2954 2955 2956 2957 2958
	/* if SCR isn't accessible on a fan-out port, PMP needs to be reset */
	if (!ata_is_host_link(link) &&
	    sata_scr_read(link, SCR_STATUS, &sstatus))
		rc = -ERESTART;

2959
	if (try >= max_tries) {
2960 2961 2962 2963 2964 2965 2966 2967
		/*
		 * Thaw host port even if reset failed, so that the port
		 * can be retried on the next phy event.  This risks
		 * repeated EH runs but seems to be a better tradeoff than
		 * shutting down a port after a botched hotplug attempt.
		 */
		if (ata_is_host_link(link))
			ata_eh_thaw_port(ap);
2968 2969
		ata_link_warn(link, "%s failed\n",
			      reset == hardreset ? "hardreset" : "softreset");
2970
		goto out;
2971
	}
2972 2973 2974 2975 2976

	now = jiffies;
	if (time_before(now, deadline)) {
		unsigned long delta = deadline - now;

2977
		ata_link_warn(failed_link,
2978 2979
			"reset failed (errno=%d), retrying in %u secs\n",
			rc, DIV_ROUND_UP(jiffies_to_msecs(delta), 1000));
2980

2981
		ata_eh_release(ap);
2982 2983
		while (delta)
			delta = schedule_timeout_uninterruptible(delta);
2984
		ata_eh_acquire(ap);
2985 2986
	}

2987 2988 2989 2990 2991 2992 2993 2994 2995 2996
	/*
	 * While disks spinup behind PMP, some controllers fail sending SRST.
	 * They need to be reset - as well as the PMP - before retrying.
	 */
	if (rc == -ERESTART) {
		if (ata_is_host_link(link))
			ata_eh_thaw_port(ap);
		goto out;
	}

Tejun Heo's avatar
Tejun Heo committed
2997
	if (try == max_tries - 1) {
2998
		sata_down_spd_limit(link, 0);
Tejun Heo's avatar
Tejun Heo committed
2999
		if (slave)
3000
			sata_down_spd_limit(slave, 0);
Tejun Heo's avatar
Tejun Heo committed
3001
	} else if (rc == -EPIPE)
3002
		sata_down_spd_limit(failed_link, 0);
Tejun Heo's avatar
Tejun Heo committed
3003

3004 3005 3006
	if (hardreset)
		reset = hardreset;
	goto retry;
3007 3008
}

3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027
static inline void ata_eh_pull_park_action(struct ata_port *ap)
{
	struct ata_link *link;
	struct ata_device *dev;
	unsigned long flags;

	/*
	 * This function can be thought of as an extended version of
	 * ata_eh_about_to_do() specially crafted to accommodate the
	 * requirements of ATA_EH_PARK handling. Since the EH thread
	 * does not leave the do {} while () loop in ata_eh_recover as
	 * long as the timeout for a park request to *one* device on
	 * the port has not expired, and since we still want to pick
	 * up park requests to other devices on the same port or
	 * timeout updates for the same device, we have to pull
	 * ATA_EH_PARK actions from eh_info into eh_context.i
	 * ourselves at the beginning of each pass over the loop.
	 *
	 * Additionally, all write accesses to &ap->park_req_pending
3028
	 * through reinit_completion() (see below) or complete_all()
3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041
	 * (see ata_scsi_park_store()) are protected by the host lock.
	 * As a result we have that park_req_pending.done is zero on
	 * exit from this function, i.e. when ATA_EH_PARK actions for
	 * *all* devices on port ap have been pulled into the
	 * respective eh_context structs. If, and only if,
	 * park_req_pending.done is non-zero by the time we reach
	 * wait_for_completion_timeout(), another ATA_EH_PARK action
	 * has been scheduled for at least one of the devices on port
	 * ap and we have to cycle over the do {} while () loop in
	 * ata_eh_recover() again.
	 */

	spin_lock_irqsave(ap->lock, flags);
3042
	reinit_completion(&ap->park_req_pending);
Tejun Heo's avatar
Tejun Heo committed
3043 3044
	ata_for_each_link(link, ap, EDGE) {
		ata_for_each_dev(dev, link, ALL) {
3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073 3074
			struct ata_eh_info *ehi = &link->eh_info;

			link->eh_context.i.dev_action[dev->devno] |=
				ehi->dev_action[dev->devno] & ATA_EH_PARK;
			ata_eh_clear_action(link, dev, ehi, ATA_EH_PARK);
		}
	}
	spin_unlock_irqrestore(ap->lock, flags);
}

static void ata_eh_park_issue_cmd(struct ata_device *dev, int park)
{
	struct ata_eh_context *ehc = &dev->link->eh_context;
	struct ata_taskfile tf;
	unsigned int err_mask;

	ata_tf_init(dev, &tf);
	if (park) {
		ehc->unloaded_mask |= 1 << dev->devno;
		tf.command = ATA_CMD_IDLEIMMEDIATE;
		tf.feature = 0x44;
		tf.lbal = 0x4c;
		tf.lbam = 0x4e;
		tf.lbah = 0x55;
	} else {
		ehc->unloaded_mask &= ~(1 << dev->devno);
		tf.command = ATA_CMD_CHK_POWER;
	}

	tf.flags |= ATA_TFLAG_DEVICE | ATA_TFLAG_ISADDR;
3075
	tf.protocol = ATA_PROT_NODATA;
3076 3077
	err_mask = ata_exec_internal(dev, &tf, NULL, DMA_NONE, NULL, 0, 0);
	if (park && (err_mask || tf.lbal != 0xc4)) {
3078
		ata_dev_err(dev, "head unload failed!\n");
3079 3080 3081 3082
		ehc->unloaded_mask &= ~(1 << dev->devno);
	}
}

3083
static int ata_eh_revalidate_and_attach(struct ata_link *link,
3084
					struct ata_device **r_failed_dev)
3085
{
3086 3087
	struct ata_port *ap = link->ap;
	struct ata_eh_context *ehc = &link->eh_context;
3088
	struct ata_device *dev;
3089
	unsigned int new_mask = 0;
3090
	unsigned long flags;
3091
	int rc = 0;
3092

3093 3094 3095 3096
	/* For PATA drive side cable detection to work, IDENTIFY must
	 * be done backwards such that PDIAG- is released by the slave
	 * device before the master device is identified.
	 */
Tejun Heo's avatar
Tejun Heo committed
3097
	ata_for_each_dev(dev, link, ALL_REVERSE) {
3098 3099
		unsigned int action = ata_eh_dev_action(dev);
		unsigned int readid_flags = 0;
3100

3101 3102 3103
		if (ehc->i.flags & ATA_EHI_DID_RESET)
			readid_flags |= ATA_READID_POSTRESET;

3104
		if ((action & ATA_EH_REVALIDATE) && ata_dev_enabled(dev)) {
3105 3106
			WARN_ON(dev->class == ATA_DEV_PMP);

3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123
			/*
			 * The link may be in a deep sleep, wake it up.
			 *
			 * If the link is in deep sleep, ata_phys_link_offline()
			 * will return true, causing the revalidation to fail,
			 * which leads to a (potentially) needless hard reset.
			 *
			 * ata_eh_recover() will later restore the link policy
			 * to ap->target_lpm_policy after revalidation is done.
			 */
			if (link->lpm_policy > ATA_LPM_MAX_POWER) {
				rc = ata_eh_set_lpm(link, ATA_LPM_MAX_POWER,
						    r_failed_dev);
				if (rc)
					goto err;
			}

Tejun Heo's avatar
Tejun Heo committed
3124
			if (ata_phys_link_offline(ata_dev_phys_link(dev))) {
3125
				rc = -EIO;
3126
				goto err;
3127 3128
			}

3129
			ata_eh_about_to_do(link, dev, ATA_EH_REVALIDATE);
3130 3131
			rc = ata_dev_revalidate(dev, ehc->classes[dev->devno],
						readid_flags);
3132
			if (rc)
3133
				goto err;
3134

3135
			ata_eh_done(link, dev, ATA_EH_REVALIDATE);
3136

3137 3138 3139 3140 3141
			/* Configuration may have changed, reconfigure
			 * transfer mode.
			 */
			ehc->i.flags |= ATA_EHI_SETMODE;

3142
			/* schedule the scsi_rescan_device() here */
3143
			schedule_delayed_work(&ap->scsi_rescan_task, 0);
3144 3145 3146
		} else if (dev->class == ATA_DEV_UNKNOWN &&
			   ehc->tries[dev->devno] &&
			   ata_class_enabled(ehc->classes[dev->devno])) {
3147 3148 3149 3150 3151 3152
			/* Temporarily set dev->class, it will be
			 * permanently set once all configurations are
			 * complete.  This is necessary because new
			 * device configuration is done in two
			 * separate loops.
			 */
3153 3154
			dev->class = ehc->classes[dev->devno];

3155 3156 3157 3158 3159
			if (dev->class == ATA_DEV_PMP)
				rc = sata_pmp_attach(dev);
			else
				rc = ata_dev_read_id(dev, &dev->class,
						     readid_flags, dev->id);
3160 3161 3162 3163 3164

			/* read_id might have changed class, store and reset */
			ehc->classes[dev->devno] = dev->class;
			dev->class = ATA_DEV_UNKNOWN;

3165 3166
			switch (rc) {
			case 0:
3167 3168
				/* clear error info accumulated during probe */
				ata_ering_clear(&dev->ering);
3169
				new_mask |= 1 << dev->devno;
3170 3171
				break;
			case -ENOENT:
3172 3173
				/* IDENTIFY was issued to non-existent
				 * device.  No need to reset.  Just
3174
				 * thaw and ignore the device.
3175 3176
				 */
				ata_eh_thaw_port(ap);
3177
				break;
3178 3179
			default:
				goto err;
3180
			}
3181 3182
		}
	}
3183

3184
	/* PDIAG- should have been released, ask cable type if post-reset */
3185 3186 3187 3188 3189
	if ((ehc->i.flags & ATA_EHI_DID_RESET) && ata_is_host_link(link)) {
		if (ap->ops->cable_detect)
			ap->cbl = ap->ops->cable_detect(ap);
		ata_force_cbl(ap);
	}
3190

3191 3192 3193
	/* Configure new devices forward such that user doesn't see
	 * device detection messages backwards.
	 */
Tejun Heo's avatar
Tejun Heo committed
3194
	ata_for_each_dev(dev, link, ALL) {
Tejun Heo's avatar
Tejun Heo committed
3195
		if (!(new_mask & (1 << dev->devno)))
3196 3197
			continue;

3198 3199
		dev->class = ehc->classes[dev->devno];

Tejun Heo's avatar
Tejun Heo committed
3200 3201 3202
		if (dev->class == ATA_DEV_PMP)
			continue;

3203 3204 3205
		ehc->i.flags |= ATA_EHI_PRINTINFO;
		rc = ata_dev_configure(dev);
		ehc->i.flags &= ~ATA_EHI_PRINTINFO;
3206 3207
		if (rc) {
			dev->class = ATA_DEV_UNKNOWN;
3208
			goto err;
3209
		}
3210 3211 3212 3213 3214 3215 3216

		spin_lock_irqsave(ap->lock, flags);
		ap->pflags |= ATA_PFLAG_SCSI_HOTPLUG;
		spin_unlock_irqrestore(ap->lock, flags);

		/* new device discovered, configure xfermode */
		ehc->i.flags |= ATA_EHI_SETMODE;
3217 3218
	}

3219
	return 0;
3220

3221
 err:
3222
	dev->flags &= ~ATA_DFLAG_RESUMING;
3223
	*r_failed_dev = dev;
3224 3225 3226
	return rc;
}

3227 3228 3229
/**
 *	ata_set_mode - Program timings and issue SET FEATURES - XFER
 *	@link: link on which timings will be programmed
3230
 *	@r_failed_dev: out parameter for failed device
3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244
 *
 *	Set ATA device disk transfer mode (PIO3, UDMA6, etc.).  If
 *	ata_set_mode() fails, pointer to the failing device is
 *	returned in @r_failed_dev.
 *
 *	LOCKING:
 *	PCI/etc. bus probe sem.
 *
 *	RETURNS:
 *	0 on success, negative errno otherwise
 */
int ata_set_mode(struct ata_link *link, struct ata_device **r_failed_dev)
{
	struct ata_port *ap = link->ap;
3245 3246
	struct ata_device *dev;
	int rc;
3247

3248
	/* if data transfer is verified, clear DUBIOUS_XFER on ering top */
Tejun Heo's avatar
Tejun Heo committed
3249
	ata_for_each_dev(dev, link, ENABLED) {
3250 3251 3252 3253 3254 3255 3256 3257 3258
		if (!(dev->flags & ATA_DFLAG_DUBIOUS_XFER)) {
			struct ata_ering_entry *ent;

			ent = ata_ering_top(&dev->ering);
			if (ent)
				ent->eflags &= ~ATA_EFLAG_DUBIOUS_XFER;
		}
	}

3259 3260
	/* has private set_mode? */
	if (ap->ops->set_mode)
3261 3262 3263 3264 3265
		rc = ap->ops->set_mode(link, r_failed_dev);
	else
		rc = ata_do_set_mode(link, r_failed_dev);

	/* if transfer mode has changed, set DUBIOUS_XFER on device */
Tejun Heo's avatar
Tejun Heo committed
3266
	ata_for_each_dev(dev, link, ENABLED) {
3267 3268 3269 3270 3271 3272 3273 3274 3275 3276
		struct ata_eh_context *ehc = &link->eh_context;
		u8 saved_xfer_mode = ehc->saved_xfer_mode[dev->devno];
		u8 saved_ncq = !!(ehc->saved_ncq_enabled & (1 << dev->devno));

		if (dev->xfer_mode != saved_xfer_mode ||
		    ata_ncq_enabled(dev) != saved_ncq)
			dev->flags |= ATA_DFLAG_DUBIOUS_XFER;
	}

	return rc;
3277 3278
}

3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297
/**
 *	atapi_eh_clear_ua - Clear ATAPI UNIT ATTENTION after reset
 *	@dev: ATAPI device to clear UA for
 *
 *	Resets and other operations can make an ATAPI device raise
 *	UNIT ATTENTION which causes the next operation to fail.  This
 *	function clears UA.
 *
 *	LOCKING:
 *	EH context (may sleep).
 *
 *	RETURNS:
 *	0 on success, -errno on failure.
 */
static int atapi_eh_clear_ua(struct ata_device *dev)
{
	int i;

	for (i = 0; i < ATA_EH_UA_TRIES; i++) {
3298
		u8 *sense_buffer = dev->sector_buf;
3299 3300 3301 3302 3303
		u8 sense_key = 0;
		unsigned int err_mask;

		err_mask = atapi_eh_tur(dev, &sense_key);
		if (err_mask != 0 && err_mask != AC_ERR_DEV) {
3304 3305 3306
			ata_dev_warn(dev,
				     "TEST_UNIT_READY failed (err_mask=0x%x)\n",
				     err_mask);
3307 3308 3309 3310 3311 3312 3313 3314
			return -EIO;
		}

		if (!err_mask || sense_key != UNIT_ATTENTION)
			return 0;

		err_mask = atapi_eh_request_sense(dev, sense_buffer, sense_key);
		if (err_mask) {
3315
			ata_dev_warn(dev, "failed to clear "
3316 3317 3318 3319 3320
				"UNIT ATTENTION (err_mask=0x%x)\n", err_mask);
			return -EIO;
		}
	}

3321 3322
	ata_dev_warn(dev, "UNIT ATTENTION persists after %d tries\n",
		     ATA_EH_UA_TRIES);
3323 3324 3325 3326

	return 0;
}

3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372
/**
 *	ata_eh_maybe_retry_flush - Retry FLUSH if necessary
 *	@dev: ATA device which may need FLUSH retry
 *
 *	If @dev failed FLUSH, it needs to be reported upper layer
 *	immediately as it means that @dev failed to remap and already
 *	lost at least a sector and further FLUSH retrials won't make
 *	any difference to the lost sector.  However, if FLUSH failed
 *	for other reasons, for example transmission error, FLUSH needs
 *	to be retried.
 *
 *	This function determines whether FLUSH failure retry is
 *	necessary and performs it if so.
 *
 *	RETURNS:
 *	0 if EH can continue, -errno if EH needs to be repeated.
 */
static int ata_eh_maybe_retry_flush(struct ata_device *dev)
{
	struct ata_link *link = dev->link;
	struct ata_port *ap = link->ap;
	struct ata_queued_cmd *qc;
	struct ata_taskfile tf;
	unsigned int err_mask;
	int rc = 0;

	/* did flush fail for this device? */
	if (!ata_tag_valid(link->active_tag))
		return 0;

	qc = __ata_qc_from_tag(ap, link->active_tag);
	if (qc->dev != dev || (qc->tf.command != ATA_CMD_FLUSH_EXT &&
			       qc->tf.command != ATA_CMD_FLUSH))
		return 0;

	/* if the device failed it, it should be reported to upper layers */
	if (qc->err_mask & AC_ERR_DEV)
		return 0;

	/* flush failed for some other reason, give it another shot */
	ata_tf_init(dev, &tf);

	tf.command = qc->tf.command;
	tf.flags |= ATA_TFLAG_DEVICE;
	tf.protocol = ATA_PROT_NODATA;

3373
	ata_dev_warn(dev, "retrying FLUSH 0x%x Emask 0x%x\n",
3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387
		       tf.command, qc->err_mask);

	err_mask = ata_exec_internal(dev, &tf, NULL, DMA_NONE, NULL, 0, 0);
	if (!err_mask) {
		/*
		 * FLUSH is complete but there's no way to
		 * successfully complete a failed command from EH.
		 * Making sure retry is allowed at least once and
		 * retrying it should do the trick - whatever was in
		 * the cache is already on the platter and this won't
		 * cause infinite loop.
		 */
		qc->scsicmd->allowed = max(qc->scsicmd->allowed, 1);
	} else {
3388
		ata_dev_warn(dev, "FLUSH failed Emask 0x%x\n",
3389 3390 3391 3392 3393 3394 3395
			       err_mask);
		rc = -EIO;

		/* if device failed it, report it to upper layers */
		if (err_mask & AC_ERR_DEV) {
			qc->err_mask |= AC_ERR_DEV;
			qc->result_tf = tf;
3396
			if (!ata_port_is_frozen(ap))
3397 3398 3399 3400 3401 3402
				rc = 0;
		}
	}
	return rc;
}

3403 3404 3405 3406 3407 3408 3409
/**
 *	ata_eh_set_lpm - configure SATA interface power management
 *	@link: link to configure power management
 *	@policy: the link power management policy
 *	@r_failed_dev: out parameter for failed device
 *
 *	Enable SATA Interface power management.  This will enable
3410 3411 3412
 *	Device Interface Power Management (DIPM) for min_power and
 *	medium_power_with_dipm policies, and then call driver specific
 *	callbacks for enabling Host Initiated Power management.
3413 3414 3415 3416 3417 3418 3419 3420 3421 3422
 *
 *	LOCKING:
 *	EH context.
 *
 *	RETURNS:
 *	0 on success, -errno on failure.
 */
static int ata_eh_set_lpm(struct ata_link *link, enum ata_lpm_policy policy,
			  struct ata_device **r_failed_dev)
{
3423
	struct ata_port *ap = ata_is_host_link(link) ? link->ap : NULL;
3424 3425
	struct ata_eh_context *ehc = &link->eh_context;
	struct ata_device *dev, *link_dev = NULL, *lpm_dev = NULL;
3426
	enum ata_lpm_policy old_policy = link->lpm_policy;
3427
	bool no_dipm = link->ap->flags & ATA_FLAG_NO_DIPM;
3428 3429 3430 3431 3432
	unsigned int hints = ATA_LPM_EMPTY | ATA_LPM_HIPM;
	unsigned int err_mask;
	int rc;

	/* if the link or host doesn't do LPM, noop */
3433 3434
	if (!IS_ENABLED(CONFIG_SATA_HOST) ||
	    (link->flags & ATA_LFLAG_NO_LPM) || (ap && !ap->ops->set_lpm))
3435 3436 3437 3438 3439 3440 3441 3442 3443 3444
		return 0;

	/*
	 * DIPM is enabled only for MIN_POWER as some devices
	 * misbehave when the host NACKs transition to SLUMBER.  Order
	 * device and link configurations such that the host always
	 * allows DIPM requests.
	 */
	ata_for_each_dev(dev, link, ENABLED) {
		bool hipm = ata_id_has_hipm(dev->id);
3445
		bool dipm = ata_id_has_dipm(dev->id) && !no_dipm;
3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458

		/* find the first enabled and LPM enabled devices */
		if (!link_dev)
			link_dev = dev;

		if (!lpm_dev && (hipm || dipm))
			lpm_dev = dev;

		hints &= ~ATA_LPM_EMPTY;
		if (!hipm)
			hints &= ~ATA_LPM_HIPM;

		/* disable DIPM before changing link config */
3459
		if (policy < ATA_LPM_MED_POWER_WITH_DIPM && dipm) {
3460 3461 3462
			err_mask = ata_dev_set_feature(dev,
					SETFEATURES_SATA_DISABLE, SATA_DIPM);
			if (err_mask && err_mask != AC_ERR_DEV) {
3463 3464 3465
				ata_dev_warn(dev,
					     "failed to disable DIPM, Emask 0x%x\n",
					     err_mask);
3466 3467 3468 3469 3470 3471
				rc = -EIO;
				goto fail;
			}
		}
	}

3472 3473 3474 3475 3476 3477
	if (ap) {
		rc = ap->ops->set_lpm(link, policy, hints);
		if (!rc && ap->slave_link)
			rc = ap->ops->set_lpm(ap->slave_link, policy, hints);
	} else
		rc = sata_pmp_set_lpm(link, policy, hints);
3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491

	/*
	 * Attribute link config failure to the first (LPM) enabled
	 * device on the link.
	 */
	if (rc) {
		if (rc == -EOPNOTSUPP) {
			link->flags |= ATA_LFLAG_NO_LPM;
			return 0;
		}
		dev = lpm_dev ? lpm_dev : link_dev;
		goto fail;
	}

3492 3493 3494 3495 3496 3497 3498 3499
	/*
	 * Low level driver acked the transition.  Issue DIPM command
	 * with the new policy set.
	 */
	link->lpm_policy = policy;
	if (ap && ap->slave_link)
		ap->slave_link->lpm_policy = policy;

3500 3501
	/* host config updated, enable DIPM if transitioning to MIN_POWER */
	ata_for_each_dev(dev, link, ENABLED) {
3502
		if (policy >= ATA_LPM_MED_POWER_WITH_DIPM && !no_dipm &&
3503
		    ata_id_has_dipm(dev->id)) {
3504 3505 3506
			err_mask = ata_dev_set_feature(dev,
					SETFEATURES_SATA_ENABLE, SATA_DIPM);
			if (err_mask && err_mask != AC_ERR_DEV) {
3507
				ata_dev_warn(dev,
3508 3509 3510 3511 3512 3513 3514 3515
					"failed to enable DIPM, Emask 0x%x\n",
					err_mask);
				rc = -EIO;
				goto fail;
			}
		}
	}

3516 3517 3518
	link->last_lpm_change = jiffies;
	link->flags |= ATA_LFLAG_CHANGED;

3519 3520 3521
	return 0;

fail:
3522 3523 3524 3525 3526
	/* restore the old policy */
	link->lpm_policy = old_policy;
	if (ap && ap->slave_link)
		ap->slave_link->lpm_policy = old_policy;

3527 3528
	/* if no device or only one more chance is left, disable LPM */
	if (!dev || ehc->tries[dev->devno] <= 2) {
3529
		ata_link_warn(link, "disabling LPM on the link\n");
3530 3531 3532 3533 3534 3535 3536
		link->flags |= ATA_LFLAG_NO_LPM;
	}
	if (r_failed_dev)
		*r_failed_dev = dev;
	return rc;
}

3537
int ata_link_nr_enabled(struct ata_link *link)
3538
{
3539 3540
	struct ata_device *dev;
	int cnt = 0;
3541

Tejun Heo's avatar
Tejun Heo committed
3542 3543
	ata_for_each_dev(dev, link, ENABLED)
		cnt++;
3544 3545 3546
	return cnt;
}

3547
static int ata_link_nr_vacant(struct ata_link *link)
3548
{
3549 3550
	struct ata_device *dev;
	int cnt = 0;
3551

Tejun Heo's avatar
Tejun Heo committed
3552
	ata_for_each_dev(dev, link, ALL)
3553
		if (dev->class == ATA_DEV_UNKNOWN)
3554 3555 3556 3557
			cnt++;
	return cnt;
}

3558
static int ata_eh_skip_recovery(struct ata_link *link)
3559
{
Tejun Heo's avatar
Tejun Heo committed
3560
	struct ata_port *ap = link->ap;
3561
	struct ata_eh_context *ehc = &link->eh_context;
3562
	struct ata_device *dev;
3563

3564 3565 3566 3567
	/* skip disabled links */
	if (link->flags & ATA_LFLAG_DISABLED)
		return 1;

3568 3569 3570 3571
	/* skip if explicitly requested */
	if (ehc->i.flags & ATA_EHI_NO_RECOVERY)
		return 1;

Tejun Heo's avatar
Tejun Heo committed
3572
	/* thaw frozen port and recover failed devices */
3573
	if (ata_port_is_frozen(ap) || ata_link_nr_enabled(link))
Tejun Heo's avatar
Tejun Heo committed
3574 3575 3576 3577 3578
		return 0;

	/* reset at least once if reset is requested */
	if ((ehc->i.action & ATA_EH_RESET) &&
	    !(ehc->i.flags & ATA_EHI_DID_RESET))
3579 3580 3581
		return 0;

	/* skip if class codes for all vacant slots are ATA_DEV_NONE */
Tejun Heo's avatar
Tejun Heo committed
3582
	ata_for_each_dev(dev, link, ALL) {
3583 3584 3585 3586 3587 3588 3589 3590
		if (dev->class == ATA_DEV_UNKNOWN &&
		    ehc->classes[dev->devno] != ATA_DEV_NONE)
			return 0;
	}

	return 1;
}

3591 3592 3593 3594 3595 3596
static int ata_count_probe_trials_cb(struct ata_ering_entry *ent, void *void_arg)
{
	u64 interval = msecs_to_jiffies(ATA_EH_PROBE_TRIAL_INTERVAL);
	u64 now = get_jiffies_64();
	int *trials = void_arg;

3597 3598
	if ((ent->eflags & ATA_EFLAG_OLD_ER) ||
	    (ent->timestamp < now - min(now, interval)))
3599 3600 3601 3602 3603 3604
		return -1;

	(*trials)++;
	return 0;
}

3605 3606 3607
static int ata_eh_schedule_probe(struct ata_device *dev)
{
	struct ata_eh_context *ehc = &dev->link->eh_context;
3608 3609
	struct ata_link *link = ata_dev_phys_link(dev);
	int trials = 0;
3610 3611 3612 3613 3614 3615 3616 3617

	if (!(ehc->i.probe_mask & (1 << dev->devno)) ||
	    (ehc->did_probe_mask & (1 << dev->devno)))
		return 0;

	ata_eh_detach_dev(dev);
	ata_dev_init(dev);
	ehc->did_probe_mask |= (1 << dev->devno);
Tejun Heo's avatar
Tejun Heo committed
3618
	ehc->i.action |= ATA_EH_RESET;
3619 3620
	ehc->saved_xfer_mode[dev->devno] = 0;
	ehc->saved_ncq_enabled &= ~(1 << dev->devno);
3621

3622
	/* the link maybe in a deep sleep, wake it up */
3623 3624 3625 3626 3627 3628 3629 3630
	if (link->lpm_policy > ATA_LPM_MAX_POWER) {
		if (ata_is_host_link(link))
			link->ap->ops->set_lpm(link, ATA_LPM_MAX_POWER,
					       ATA_LPM_EMPTY);
		else
			sata_pmp_set_lpm(link, ATA_LPM_MAX_POWER,
					 ATA_LPM_EMPTY);
	}
3631

3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650
	/* Record and count probe trials on the ering.  The specific
	 * error mask used is irrelevant.  Because a successful device
	 * detection clears the ering, this count accumulates only if
	 * there are consecutive failed probes.
	 *
	 * If the count is equal to or higher than ATA_EH_PROBE_TRIALS
	 * in the last ATA_EH_PROBE_TRIAL_INTERVAL, link speed is
	 * forced to 1.5Gbps.
	 *
	 * This is to work around cases where failed link speed
	 * negotiation results in device misdetection leading to
	 * infinite DEVXCHG or PHRDY CHG events.
	 */
	ata_ering_record(&dev->ering, 0, AC_ERR_OTHER);
	ata_ering_map(&dev->ering, ata_count_probe_trials_cb, &trials);

	if (trials > ATA_EH_PROBE_TRIALS)
		sata_down_spd_limit(link, 1);

3651 3652 3653
	return 1;
}

3654
static int ata_eh_handle_dev_fail(struct ata_device *dev, int err)
3655
{
Tejun Heo's avatar
Tejun Heo committed
3656
	struct ata_eh_context *ehc = &dev->link->eh_context;
3657

3658 3659 3660 3661 3662
	/* -EAGAIN from EH routine indicates retry without prejudice.
	 * The requester is responsible for ensuring forward progress.
	 */
	if (err != -EAGAIN)
		ehc->tries[dev->devno]--;
3663 3664 3665 3666 3667

	switch (err) {
	case -ENODEV:
		/* device missing or wrong IDENTIFY data, schedule probing */
		ehc->i.probe_mask |= (1 << dev->devno);
3668
		fallthrough;
3669 3670 3671
	case -EINVAL:
		/* give it just one more chance */
		ehc->tries[dev->devno] = min(ehc->tries[dev->devno], 1);
3672
		fallthrough;
3673
	case -EIO:
3674
		if (ehc->tries[dev->devno] == 1) {
3675 3676 3677
			/* This is the last chance, better to slow
			 * down than lose it.
			 */
3678
			sata_down_spd_limit(ata_dev_phys_link(dev), 0);
3679 3680
			if (dev->pio_mode > XFER_PIO_0)
				ata_down_xfermask_limit(dev, ATA_DNXFER_PIO);
3681 3682 3683 3684 3685 3686 3687 3688
		}
	}

	if (ata_dev_enabled(dev) && !ehc->tries[dev->devno]) {
		/* disable device if it has used up all its chances */
		ata_dev_disable(dev);

		/* detach if offline */
Tejun Heo's avatar
Tejun Heo committed
3689
		if (ata_phys_link_offline(ata_dev_phys_link(dev)))
3690 3691
			ata_eh_detach_dev(dev);

3692
		/* schedule probe if necessary */
3693
		if (ata_eh_schedule_probe(dev)) {
3694
			ehc->tries[dev->devno] = ATA_EH_DEV_TRIES;
3695 3696 3697
			memset(ehc->cmd_timeout_idx[dev->devno], 0,
			       sizeof(ehc->cmd_timeout_idx[dev->devno]));
		}
3698 3699

		return 1;
3700
	} else {
Tejun Heo's avatar
Tejun Heo committed
3701
		ehc->i.action |= ATA_EH_RESET;
3702
		return 0;
3703 3704 3705
	}
}

3706 3707 3708
/**
 *	ata_eh_recover - recover host port after error
 *	@ap: host port to recover
3709
 *	@prereset: prereset method (can be NULL)
3710 3711 3712
 *	@softreset: softreset method (can be NULL)
 *	@hardreset: hardreset method (can be NULL)
 *	@postreset: postreset method (can be NULL)
3713
 *	@r_failed_link: out parameter for failed link
3714 3715 3716
 *
 *	This is the alpha and omega, eum and yang, heart and soul of
 *	libata exception handling.  On entry, actions required to
3717 3718 3719
 *	recover each link and hotplug requests are recorded in the
 *	link's eh_context.  This function executes all the operations
 *	with appropriate retrials and fallbacks to resurrect failed
3720
 *	devices, detach goners and greet newcomers.
3721 3722 3723 3724 3725 3726 3727
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 *
 *	RETURNS:
 *	0 on success, -errno on failure.
 */
3728 3729 3730 3731
int ata_eh_recover(struct ata_port *ap, ata_prereset_fn_t prereset,
		   ata_reset_fn_t softreset, ata_reset_fn_t hardreset,
		   ata_postreset_fn_t postreset,
		   struct ata_link **r_failed_link)
3732
{
3733
	struct ata_link *link;
3734
	struct ata_device *dev;
3735
	int rc, nr_fails;
3736
	unsigned long flags, deadline;
3737 3738

	/* prep for recovery */
Tejun Heo's avatar
Tejun Heo committed
3739
	ata_for_each_link(link, ap, EDGE) {
3740
		struct ata_eh_context *ehc = &link->eh_context;
3741

3742 3743 3744 3745 3746 3747 3748 3749 3750
		/* re-enable link? */
		if (ehc->i.action & ATA_EH_ENABLE_LINK) {
			ata_eh_about_to_do(link, NULL, ATA_EH_ENABLE_LINK);
			spin_lock_irqsave(ap->lock, flags);
			link->flags &= ~ATA_LFLAG_DISABLED;
			spin_unlock_irqrestore(ap->lock, flags);
			ata_eh_done(link, NULL, ATA_EH_ENABLE_LINK);
		}

Tejun Heo's avatar
Tejun Heo committed
3751
		ata_for_each_dev(dev, link, ALL) {
3752 3753 3754 3755
			if (link->flags & ATA_LFLAG_NO_RETRY)
				ehc->tries[dev->devno] = 1;
			else
				ehc->tries[dev->devno] = ATA_EH_DEV_TRIES;
3756

3757 3758 3759 3760 3761 3762 3763 3764 3765
			/* collect port action mask recorded in dev actions */
			ehc->i.action |= ehc->i.dev_action[dev->devno] &
					 ~ATA_EH_PERDEV_MASK;
			ehc->i.dev_action[dev->devno] &= ATA_EH_PERDEV_MASK;

			/* process hotplug request */
			if (dev->flags & ATA_DFLAG_DETACH)
				ata_eh_detach_dev(dev);

3766 3767 3768
			/* schedule probe if necessary */
			if (!ata_dev_enabled(dev))
				ata_eh_schedule_probe(dev);
3769
		}
3770 3771 3772 3773 3774
	}

 retry:
	rc = 0;

3775
	/* if UNLOADING, finish immediately */
3776
	if (ap->pflags & ATA_PFLAG_UNLOADING)
3777 3778
		goto out;

3779
	/* prep for EH */
Tejun Heo's avatar
Tejun Heo committed
3780
	ata_for_each_link(link, ap, EDGE) {
3781
		struct ata_eh_context *ehc = &link->eh_context;
3782

3783 3784 3785 3786
		/* skip EH if possible. */
		if (ata_eh_skip_recovery(link))
			ehc->i.action = 0;

Tejun Heo's avatar
Tejun Heo committed
3787
		ata_for_each_dev(dev, link, ALL)
3788 3789
			ehc->classes[dev->devno] = ATA_DEV_UNKNOWN;
	}
3790

3791
	/* reset */
Tejun Heo's avatar
Tejun Heo committed
3792
	ata_for_each_link(link, ap, EDGE) {
3793
		struct ata_eh_context *ehc = &link->eh_context;
3794

3795 3796
		if (!(ehc->i.action & ATA_EH_RESET))
			continue;
3797

3798 3799 3800
		rc = ata_eh_reset(link, ata_link_nr_vacant(link),
				  prereset, softreset, hardreset, postreset);
		if (rc) {
3801
			ata_link_err(link, "reset failed, giving up\n");
3802
			goto out;
3803 3804 3805
		}
	}

3806 3807 3808 3809 3810 3811 3812 3813 3814 3815
	do {
		unsigned long now;

		/*
		 * clears ATA_EH_PARK in eh_info and resets
		 * ap->park_req_pending
		 */
		ata_eh_pull_park_action(ap);

		deadline = jiffies;
Tejun Heo's avatar
Tejun Heo committed
3816 3817
		ata_for_each_link(link, ap, EDGE) {
			ata_for_each_dev(dev, link, ALL) {
3818 3819 3820
				struct ata_eh_context *ehc = &link->eh_context;
				unsigned long tmp;

3821 3822
				if (dev->class != ATA_DEV_ATA &&
				    dev->class != ATA_DEV_ZAC)
3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842
					continue;
				if (!(ehc->i.dev_action[dev->devno] &
				      ATA_EH_PARK))
					continue;
				tmp = dev->unpark_deadline;
				if (time_before(deadline, tmp))
					deadline = tmp;
				else if (time_before_eq(tmp, jiffies))
					continue;
				if (ehc->unloaded_mask & (1 << dev->devno))
					continue;

				ata_eh_park_issue_cmd(dev, 1);
			}
		}

		now = jiffies;
		if (time_before_eq(deadline, now))
			break;

3843
		ata_eh_release(ap);
3844 3845
		deadline = wait_for_completion_timeout(&ap->park_req_pending,
						       deadline - now);
3846
		ata_eh_acquire(ap);
3847
	} while (deadline);
Tejun Heo's avatar
Tejun Heo committed
3848 3849
	ata_for_each_link(link, ap, EDGE) {
		ata_for_each_dev(dev, link, ALL) {
3850 3851 3852 3853 3854 3855 3856 3857 3858
			if (!(link->eh_context.unloaded_mask &
			      (1 << dev->devno)))
				continue;

			ata_eh_park_issue_cmd(dev, 0);
			ata_eh_done(link, dev, ATA_EH_PARK);
		}
	}

3859
	/* the rest */
3860 3861
	nr_fails = 0;
	ata_for_each_link(link, ap, PMP_FIRST) {
3862
		struct ata_eh_context *ehc = &link->eh_context;
3863

3864 3865 3866
		if (sata_pmp_attached(ap) && ata_is_host_link(link))
			goto config_lpm;

3867 3868
		/* revalidate existing devices and attach new ones */
		rc = ata_eh_revalidate_and_attach(link, &dev);
3869
		if (rc)
3870
			goto rest_fail;
3871

3872 3873 3874 3875 3876 3877
		/* if PMP got attached, return, pmp EH will take care of it */
		if (link->device->class == ATA_DEV_PMP) {
			ehc->i.action = 0;
			return 0;
		}

3878 3879 3880 3881
		/* configure transfer mode if necessary */
		if (ehc->i.flags & ATA_EHI_SETMODE) {
			rc = ata_set_mode(link, &dev);
			if (rc)
3882
				goto rest_fail;
3883 3884 3885
			ehc->i.flags &= ~ATA_EHI_SETMODE;
		}

3886 3887 3888 3889
		/* If reset has been issued, clear UA to avoid
		 * disrupting the current users of the device.
		 */
		if (ehc->i.flags & ATA_EHI_DID_RESET) {
Tejun Heo's avatar
Tejun Heo committed
3890
			ata_for_each_dev(dev, link, ALL) {
3891 3892 3893 3894
				if (dev->class != ATA_DEV_ATAPI)
					continue;
				rc = atapi_eh_clear_ua(dev);
				if (rc)
3895
					goto rest_fail;
3896 3897
				if (zpodd_dev_enabled(dev))
					zpodd_post_poweron(dev);
3898 3899 3900
			}
		}

3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911
		/*
		 * Make sure to transition devices to the active power mode
		 * if needed (e.g. if we were scheduled on system resume).
		 */
		ata_for_each_dev(dev, link, ENABLED) {
			if (ehc->i.dev_action[dev->devno] & ATA_EH_SET_ACTIVE) {
				ata_dev_power_set_active(dev);
				ata_eh_done(link, dev, ATA_EH_SET_ACTIVE);
			}
		}

3912 3913
		/* retry flush if necessary */
		ata_for_each_dev(dev, link, ALL) {
3914 3915
			if (dev->class != ATA_DEV_ATA &&
			    dev->class != ATA_DEV_ZAC)
3916 3917 3918
				continue;
			rc = ata_eh_maybe_retry_flush(dev);
			if (rc)
3919
				goto rest_fail;
3920 3921
		}

3922
	config_lpm:
3923
		/* configure link power saving */
3924 3925 3926 3927 3928
		if (link->lpm_policy != ap->target_lpm_policy) {
			rc = ata_eh_set_lpm(link, ap->target_lpm_policy, &dev);
			if (rc)
				goto rest_fail;
		}
3929

3930 3931 3932
		/* this link is okay now */
		ehc->i.flags = 0;
		continue;
3933

3934 3935 3936 3937
	rest_fail:
		nr_fails++;
		if (dev)
			ata_eh_handle_dev_fail(dev, rc);
3938

3939
		if (ata_port_is_frozen(ap)) {
3940 3941 3942
			/* PMP reset requires working host port.
			 * Can't retry if it's frozen.
			 */
Tejun Heo's avatar
Tejun Heo committed
3943
			if (sata_pmp_attached(ap))
3944
				goto out;
3945
			break;
3946
		}
3947 3948
	}

3949
	if (nr_fails)
3950
		goto retry;
3951

3952 3953 3954 3955
 out:
	if (rc && r_failed_link)
		*r_failed_link = link;

3956 3957 3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968
	return rc;
}

/**
 *	ata_eh_finish - finish up EH
 *	@ap: host port to finish EH for
 *
 *	Recovery is complete.  Clean up EH states and retry or finish
 *	failed qcs.
 *
 *	LOCKING:
 *	None.
 */
3969
void ata_eh_finish(struct ata_port *ap)
3970
{
3971
	struct ata_queued_cmd *qc;
3972 3973 3974
	int tag;

	/* retry or finish qcs */
3975
	ata_qc_for_each_raw(ap, qc, tag) {
3976
		if (!(qc->flags & ATA_QCFLAG_EH))
3977 3978 3979 3980 3981 3982 3983
			continue;

		if (qc->err_mask) {
			/* FIXME: Once EH migration is complete,
			 * generate sense data in this function,
			 * considering both err_mask and tf.
			 */
3984 3985 3986 3987 3988 3989 3990
			if (qc->flags & ATA_QCFLAG_RETRY) {
				/*
				 * Since qc->err_mask is set, ata_eh_qc_retry()
				 * will not increment scmd->allowed, so upper
				 * layer will only retry the command if it has
				 * not already been retried too many times.
				 */
3991
				ata_eh_qc_retry(qc);
3992
			} else {
3993
				ata_eh_qc_complete(qc);
3994
			}
3995
		} else {
3996 3997
			if (qc->flags & ATA_QCFLAG_SENSE_VALID ||
			    qc->flags & ATA_QCFLAG_EH_SUCCESS_CMD) {
3998 3999 4000 4001
				ata_eh_qc_complete(qc);
			} else {
				/* feed zero TF to sense generation */
				memset(&qc->result_tf, 0, sizeof(qc->result_tf));
4002 4003 4004 4005 4006 4007
				/*
				 * Since qc->err_mask is not set,
				 * ata_eh_qc_retry() will increment
				 * scmd->allowed, so upper layer is guaranteed
				 * to retry the command.
				 */
4008 4009 4010 4011
				ata_eh_qc_retry(qc);
			}
		}
	}
4012 4013 4014 4015

	/* make sure nr_active_links is zero after EH */
	WARN_ON(ap->nr_active_links);
	ap->nr_active_links = 0;
4016 4017 4018 4019 4020
}

/**
 *	ata_do_eh - do standard error handling
 *	@ap: host port to handle error for
4021
 *
4022
 *	@prereset: prereset method (can be NULL)
4023 4024 4025 4026 4027 4028 4029 4030 4031
 *	@softreset: softreset method (can be NULL)
 *	@hardreset: hardreset method (can be NULL)
 *	@postreset: postreset method (can be NULL)
 *
 *	Perform standard error handling sequence.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 */
4032 4033 4034
void ata_do_eh(struct ata_port *ap, ata_prereset_fn_t prereset,
	       ata_reset_fn_t softreset, ata_reset_fn_t hardreset,
	       ata_postreset_fn_t postreset)
4035
{
4036 4037 4038 4039 4040 4041 4042 4043 4044
	struct ata_device *dev;
	int rc;

	ata_eh_autopsy(ap);
	ata_eh_report(ap);

	rc = ata_eh_recover(ap, prereset, softreset, hardreset, postreset,
			    NULL);
	if (rc) {
Tejun Heo's avatar
Tejun Heo committed
4045
		ata_for_each_dev(dev, &ap->link, ALL)
4046 4047 4048
			ata_dev_disable(dev);
	}

4049 4050
	ata_eh_finish(ap);
}
4051

4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064 4065
/**
 *	ata_std_error_handler - standard error handler
 *	@ap: host port to handle error for
 *
 *	Standard error handler
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 */
void ata_std_error_handler(struct ata_port *ap)
{
	struct ata_port_operations *ops = ap->ops;
	ata_reset_fn_t hardreset = ops->hardreset;

4066
	/* ignore built-in hardreset if SCR access is not available */
4067
	if (hardreset == sata_std_hardreset && !sata_scr_valid(&ap->link))
4068 4069 4070 4071
		hardreset = NULL;

	ata_do_eh(ap, ops->prereset, ops->softreset, hardreset, ops->postreset);
}
4072
EXPORT_SYMBOL_GPL(ata_std_error_handler);
4073

4074
#ifdef CONFIG_PM
4075 4076 4077 4078 4079 4080 4081 4082 4083 4084 4085 4086 4087
/**
 *	ata_eh_handle_port_suspend - perform port suspend operation
 *	@ap: port to suspend
 *
 *	Suspend @ap.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 */
static void ata_eh_handle_port_suspend(struct ata_port *ap)
{
	unsigned long flags;
	int rc = 0;
4088
	struct ata_device *dev;
4089
	struct ata_link *link;
4090 4091 4092 4093

	/* are we suspending? */
	spin_lock_irqsave(ap->lock, flags);
	if (!(ap->pflags & ATA_PFLAG_PM_PENDING) ||
4094
	    ap->pm_mesg.event & PM_EVENT_RESUME) {
4095 4096 4097 4098 4099 4100 4101
		spin_unlock_irqrestore(ap->lock, flags);
		return;
	}
	spin_unlock_irqrestore(ap->lock, flags);

	WARN_ON(ap->pflags & ATA_PFLAG_SUSPENDED);

4102 4103 4104 4105 4106 4107
	/* Set all devices attached to the port in standby mode */
	ata_for_each_link(link, ap, HOST_FIRST) {
		ata_for_each_dev(dev, link, ENABLED)
			ata_dev_power_set_standby(dev);
	}

4108 4109 4110
	/*
	 * If we have a ZPODD attached, check its zero
	 * power ready status before the port is frozen.
4111
	 * Only needed for runtime suspend.
4112
	 */
4113 4114 4115 4116 4117
	if (PMSG_IS_AUTO(ap->pm_mesg)) {
		ata_for_each_dev(dev, &ap->link, ENABLED) {
			if (zpodd_dev_enabled(dev))
				zpodd_on_suspend(dev);
		}
4118 4119
	}

4120 4121 4122 4123 4124 4125
	/* suspend */
	ata_eh_freeze_port(ap);

	if (ap->ops->port_suspend)
		rc = ap->ops->port_suspend(ap, ap->pm_mesg);

4126
	ata_acpi_set_state(ap, ap->pm_mesg);
4127

4128
	/* update the flags */
4129 4130 4131 4132 4133
	spin_lock_irqsave(ap->lock, flags);

	ap->pflags &= ~ATA_PFLAG_PM_PENDING;
	if (rc == 0)
		ap->pflags |= ATA_PFLAG_SUSPENDED;
4134
	else if (ata_port_is_frozen(ap))
4135 4136 4137 4138 4139 4140 4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152
		ata_port_schedule_eh(ap);

	spin_unlock_irqrestore(ap->lock, flags);

	return;
}

/**
 *	ata_eh_handle_port_resume - perform port resume operation
 *	@ap: port to resume
 *
 *	Resume @ap.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 */
static void ata_eh_handle_port_resume(struct ata_port *ap)
{
Tejun Heo's avatar
Tejun Heo committed
4153 4154
	struct ata_link *link;
	struct ata_device *dev;
4155 4156 4157 4158 4159
	unsigned long flags;

	/* are we resuming? */
	spin_lock_irqsave(ap->lock, flags);
	if (!(ap->pflags & ATA_PFLAG_PM_PENDING) ||
4160
	    !(ap->pm_mesg.event & PM_EVENT_RESUME)) {
4161 4162 4163 4164 4165
		spin_unlock_irqrestore(ap->lock, flags);
		return;
	}
	spin_unlock_irqrestore(ap->lock, flags);

4166
	WARN_ON(!(ap->pflags & ATA_PFLAG_SUSPENDED));
4167

Tejun Heo's avatar
Tejun Heo committed
4168 4169 4170 4171 4172 4173 4174 4175 4176 4177 4178
	/*
	 * Error timestamps are in jiffies which doesn't run while
	 * suspended and PHY events during resume isn't too uncommon.
	 * When the two are combined, it can lead to unnecessary speed
	 * downs if the machine is suspended and resumed repeatedly.
	 * Clear error history.
	 */
	ata_for_each_link(link, ap, HOST_FIRST)
		ata_for_each_dev(dev, link, ALL)
			ata_ering_clear(&dev->ering);

4179
	ata_acpi_set_state(ap, ap->pm_mesg);
4180

4181
	if (ap->ops->port_resume)
4182
		ap->ops->port_resume(ap);
4183

4184 4185 4186
	/* tell ACPI that we're resuming */
	ata_acpi_on_resume(ap);

4187
	/* update the flags */
4188 4189
	spin_lock_irqsave(ap->lock, flags);
	ap->pflags &= ~(ATA_PFLAG_PM_PENDING | ATA_PFLAG_SUSPENDED);
4190
	ap->pflags |= ATA_PFLAG_RESUMING;
4191 4192
	spin_unlock_irqrestore(ap->lock, flags);
}
4193
#endif /* CONFIG_PM */